article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
x - ray observations are playing an increasingly important role in the study of galaxies . with its arcsecond spatial resolution ,
in particular has made a significant impact on our understanding of discrete x - ray source populations , which mostly represent various stellar end - products [ e.g. , low- and high - mass x - ray binaries ( lmxbs and hmxbs ) and supernova remnants ( snrs ) ] as well as active galactic nuclei ( agns ) .
the resolution also allows for a clean excision of such sources from the data so low - surface brightness emission ( e.g. from diffuse hot gas ) can be mapped out . an under - appreciated aspect of is its spectroscopic capability in the study of diffuse hot gas when the low- and high - energy grating instruments are used . although the sensitivities of the instruments are quite limited , the existing observations of a dozen or so bright objects ( agns and lmxbs ) have yielded data of high enough quality for unprecedented x - ray absorption line spectroscopic measurements of the global hot gas in and around our galaxy .
useful constraints have also been obtained on the overall content of hot gas around galaxies within certain impact distances of the sight lines toward the agns .
these measurements , compared with physical models and simulations of the hot gas , are shedding important insights on its relationship to the feedback from stars and agns .
these aspects of the s legacy are reviewed in the following .
the overall x - ray luminosity of a galaxy ( except for a giant elliptical ) is usually dominated by hmxbs and/or lmxbs .
the luminosity functions of hmxbs and lmxbs are linearly scaled with the star formation rate and the stellar mass of a galaxy and are universal to an accuracy better than @xmath0 and @xmath1 , respectively @xcite .
the differential power law slope of the function for hmxbs is @xmath2 over a broad range of log@xmath3 , where @xmath4 is the luminosity in the 0.5 - 2 kev band .
particularly interesting are a large number of non - agn ( hyper)ultraluminous x - ray sources with log@xmath5 in the range of @xmath6 and observed typically in active star forming galaxies , suggesting the presence of either so - called intermediate - mass black holes ( @xmath7 ) or sources apparently radiating well above the eddington limit . the luminosity function shape for lmxbs is a bit more complicated , having a slope of @xmath8 at log@xmath9 , steepening gradually at higher luminosities and cutting off abruptly at log@xmath10 .
furthermore , the frequency of lmxbs per stellar mass is substantially higher in globular clusters than in galaxy field ( e.g. , @xcite ) .
this is attributed to the formation of lmxbs via close stellar encounters , which has also been proposed to account for an enhanced number of lmxbs in the dense inner bulge of m 31 @xcite .
but it is not yet clear as to what fraction of field lmxbs was formed dynamically ( e.g. , @xcite ) . for the study of diffuse hot gas
, it is important to minimize the confusion from point - like source contributions .
a source detection limit at least down to log@xmath11 is highly desirable , which can be achieved with a _
observation of a reasonably deep exposure for nearby galaxies ( @xmath12 mpc ) .
the residual contribution from fainter hmxbs and lmxbs can then be estimated from their correlation with the star formation rate and stellar mass and subtracted from the data with little uncertainty
. however , one still needs to be careful with poisson fluctuations of sources just below the detection limit .
such fluctuations may significantly affect the reliability of x - ray morphological analysis of a galaxy .
in addition to the subtraction of relatively bright x - ray binaries , one also needs to account for a significant ( even dominant ) stellar contribution from cataclysmic variables and coronally active stars , which are numerous , though individually faint .
very deep imaging of a region toward the galactic bulge @xcite has resolved out more than 80% of the background emission at energies of 6 - 7 kev , where the observed prominent fe @xmath13-kev line was thought as the evidence for the presence of diffuse hot plasma at @xmath14 k. this high - energy background emission is now shown to be entirely consistent with this collective stellar contribution in the galactic bulge / ridge .
it should be noted , however , that the resolved fraction is much smaller at lower energies ( @xmath0 at @xmath15 kev ) , which may be considered as an indication for the presence of diffuse hot gas at much lower temperatures .
the stellar contribution is unresolved for external galaxies , even nearby ones .
fortunately , the average x - ray spectrum and specific luminosity ( per stellar mass ) of the contribution have been calibrated , based on the observations of m32 , which is too light to hold significant amount of diffuse hot gas ) , together with the direct detection of stellar x - ray sources in the solar neighborhood @xcite .
the contribution can be readily included in a spectral analysis of the `` diffuse '' x - ray emission of a galaxy . in an imaging analysis
, one may subtract the contribution scaled according to the stellar mass distribution ( e.g. , traced by the near - ir k - band intensity of a galaxy ) .
lines of key species that trace gas in the temperature range of @xmath16 k @xcite .
, scaledwidth=50.0% ]
the properties of the global hot gas on scales comparable to the size of the galaxy remained largely unknown until recently .
before , we did have various broad - band x - ray emission surveys of the sky such as the one made by in the 0.1 - 2.4 kev range , which is sensitive to the hot gas @xcite .
but such a survey alone can not be used to directly determine the origin of the x - ray emission , which carries little distance information .
the interpretation of the emission depends sensitively on the assumed cool ( x - ray - absorbing ) and hot gas distributions .
spectroscopic information on the x - ray emission has been obtained from rocket experiments @xcite and more recently from observations , but only for a number of sample regions ( e.g. , @xcite ) .
there are also large uncertainties in the contributions from faint discrete sources and other irrelevant phenomena such as solar wind charge exchange ( swcx ) to the emission .
a breakthrough for the study of the global hot gas has been made from the use of the x - ray absorption line spectroscopy ( e.g. , @xcite ) .
while absorption line spectroscopy is commonly used in optical and uv studies of the interstellar medium ( ism ) , this technique became feasible in the soft x - ray regime only with grating spectra from and .
unlike the emission , which is sensitive to the density structure , absorption lines produced by ions such as o vii , o viii , and ne ix ( fig . 1 [ fig : f1 ] ) directly probe their column densities , which are proportional to the mass of the hot gas .
the relative strengths of such absorption lines give direct diagnostics of the thermal , chemical and/or kinetic properties of the hot gas .
although the absorption lines are rarely resolved in the spectra ( with a resolution of @xmath17 fwhm ) , the velocity dispersion of the hot gas can be derived from the relative line saturation of different transitions of same ion species ( e.g. , k@xmath18 vs. k@xmath19 ) . because the k - shell transitions of carbon through iron are all in the x - ray regime
, the same technique can be used to study the ism in essentially all phases ( cold , warm , and hot ) and forms ( atomic , molecular , and dust grain ; @xcite ) .
furthermore , the measurements are insensitive to the photo - electric absorption by the cool ism ( @xmath20 k ) and to the swcx
. therefore , the x - ray absorption line spectroscopy allows us to probe the global ism unbiasedly along a sight line .
the effectiveness of the technique can be further enhanced when multiple sight lines are analyzed jointly ( e.g. , @xcite ) and/or emission data are included ( e.g. , @xcite ) .
we can then infer differential properties of hot gas between sight lines of different depths or directions and/or estimate the pathlength and density of the hot gas .
the application of this x - ray tomography , though only to a very limited number of sight lines so far , has led to the first characterization of the global hot gas : * the spatial distribution of the gas at the solar neighborhood can be characterized by a thick galactic disk with a density scale height of @xmath21 kpc ( e.g. , @xcite ) , comparable to those measured for ( from far - uv absorption line observations ) and free electrons ( from pulsar dispersion measures ) .
the density is enhanced toward the inner region of the galaxy , indicating the presence of a galactic bulge component of the hot gas @xcite .
a temperature increase is also observed from the disk ( @xmath22 k ) to the bulge ( @xmath23 k ; e.g. , @xcite ) . * the velocity dispersion increases from @xmath24 at the solar neighborhood to @xmath25 toward the inner region of the galaxy ( with 90% uncertainties up to @xmath26 ; e.g. , @xcite ) , indicating a significant nonthermal ( e.g. , turbulent ) velocity contribution to the line broadening ( cf .
the thermal broadening @xmath27 ) . * the metal abundances ( o / fe , fe / ne , and o / ne ) are about solar , although there are lines of evidence for the depletion of o and fe ( by a factor of @xmath21 ) at the solar neighborhood ( e.g. , @xcite ) , assuming the abundances in @xcite . .
* no evidence is yet found for a large - scale x - ray - absorbing hot circum - galactic medium ( cgm ) with a 95% upper limit to the column density @xmath28 for regions @xmath29 kpc away from the galactic disk / bulge @xcite and @xmath30 for @xmath31 kpc @xcite .
this low column density of the hot cgm or the local hot intragroup medium is consistent with the lack of evidence for enhanced x - ray line absorptions along sight lines near the major axis of the local group @xcite .
this basic characterization demonstrates the potential of using the x - ray tomography to greatly advance our understanding of the global hot gas in and around the galaxy .
the x - ray absorption line spectroscopy can also be used to probe the hot cgm around other galaxies .
yao et al . have recently studied the hot cgm along the sight lines toward luminous agns @xcite . as an example , fig .
2 shows a grating spectrum of pks 2155 - 304 .
no absorption line is apparent at the redshifts of the three groups of galaxies with the impact distances smaller than 500 kpc of the sight line ( in particular , the group at the highest redshift contains two galaxies brighter than @xmath32 ) .
the same is true when the spectrum is folded according to the redshifts of the individual groups . to increase the counting statistics further , yao et al .
have stacked spectra from the grating observations of eight agns with such intervening galaxies or groups .
no significant absorption is detected for any of these individual systems or in the final stacked spectrum with a total equivalent exposure of about 10 ms !
upper limits to the mean column densities of various ion species per galaxy or group are then estimated .
in particular , they find that @xmath33 , consistent with the constraints for the cgm around our galaxy .
they have estimated the total mass contained in the cgm as @xmath34 , where @xmath35 , @xmath36 , and @xmath37 are the ionization fraction of , metal abundance , and the radius of the hot cgm , respectively .
this is in contrast to the expected baryon mass @xmath38 for the halo of a milky way - type galaxy or a typical galaxy group @xcite .
thus the bulk of the cgm unlikely resides in such a chemically enriched warm - hot phase at temperatures ranging from @xmath16 k ( fig . 1 [ fig : f1 ] ) , which our x - ray absorption line spectroscopy is sensitive to .
this conclusion has strong implications for understanding the accumulated effect of the stellar and agn feedback on the galactic ecosystem ( see the discussion section ) . to study the effect of ongoing stellar and agn feedback
, one can map out diffuse x - ray emission from hot gas in and around nearby galaxies of various masses and star formation rates .
much attention has been paid to the feedback in starburst and massive elliptical galaxies , which are relatively bright in diffuse x - ray emission .
observations have shown convincingly that the agn feedback is important in shaping the morphology and thermal evolution of hot gas in massive elliptical galaxies , particularly those at centers of galaxy groups and clusters ( @xcite and references therein ) .
the asymmetry in the global diffuse x - ray morphology is correlated with radio and x - ray luminosities of agns in elliptical galaxies , even in rather x - ray - faint ones @xcite .
this calls into question the hydrostatic assumption commonly used in order to infer the gravitational mass distribution in such galaxies .
nevertheless , the hydrostatic assumption may hold approximately for hot gas around the central supermassive black holes ( smbhs ) , if they are in a sufficiently quiescent state .
the smbh masses may then be measured from spatially resolved x - ray spectroscopy of the hot gas .
humphrey et al . have made such mass measurements for four smbhs with data @xcite .
three of them already have mass determinations from the kinematics of either stars or a central gas disk .
it is encouraging to find a good agreement between the measurements using the different methods . from this agreement
, they further infer that no more than @xmath39 of the ism pressure around the smbhs should be nonthermal .
the feedback in nuclear starburst galaxies is manifested in the so - called galactic superwinds driven by the mechanical energy injection from fast stellar winds and supernovae ( sne ) of massive stars ( e.g. , @xcite ) . the observed soft x - ray emission from a superwind typically has an elongated morphology along the minor axis of such a galaxy and is correlated well with extraplanar h@xmath18-emitting features .
this indicates that the detected hot gas arises primarily from the interaction between the superwind and cool gas .
the superwind itself , believed to be very hot and low in density , is much difficult to detect . from a detailed comparison between data and hydrodynamic simulations , strickland & heckman
infer that the superwind of m82 has a mean temperature of @xmath40 k and a mass outflowing rate of @xmath41 @xcite .
such energetic superwinds with little radiative energy loss must have profound effects on the large - scale cgm ( e.g. , @xcite ) .
recent x - ray observations have further shown the importance of the feedback in understanding even `` normal '' intermediate - mass galaxies ( similar to the milky way and m31 ; e.g. , @xcite ) . , in particular , has unambiguously detected diffuse hot gas in and around normal disk galaxies .
the total x - ray luminosity of the gas is well correlated with the star formation rate for such galaxies .
the diffuse soft x - ray emission is shown to be strongly enhanced in recent star forming regions or spiral arms within an individual galaxy viewed face - on and is only slightly more diffuse than h@xmath18 emission ( e.g. , @xcite ) .
this narrow appearance of spiral arms in x - ray conflicts the expectation from population synthesis models : the mechanical energy output rate from sne should be nearly constant over a time period that is up to 10 times longer than the lifetime of massive ionizing stars .
this means that sne in inter - arm regions , where ism density is generally low , must produce less soft x - ray emission than those in the arms . when disk galaxies are observed in an inclined angle ( e.g. , fig .
3 ) , the soft x - ray emission tends to appear as plumes , most likely representing blown - out hot gas heated in recent massive star forming regions and galactic spheroids .
the cleanest perspectives of the extraplanar hot gas are obtained from the observations of edge - on disk galaxies such as ngc 891 @xcite and ngc 5775 @xcite .
the observed diffuse x - ray emission typically does not extend significantly more than a few kpc away from the galactic disks . a claimed detection of the emission around the edge - on spiral ngc 5746 on larger scales was later proved to be due to an instrumental artifact @xcite .
complementary observations from _ xmm - newton _ and _ suzaku _ give consistent results and provide improved spectral information on the hot gas ( e.g. , @xcite ) .
the morphology of the x - ray emission as well as its correlation with the star formation rate clearly shows that the extraplanar hot gas is primarily heated by the stellar feedback .
in fact , the cooling of the hot gas accounts for only a small fraction of the expected energy input from massive stars alone ( typically no more than a few % ) .
the best example of mapping out hot gas in a galactic stellar bulge ( or spheroid ) is the discovery of an apparent hot gas outflow from the m31 bulge @xcite .
this outflow with a 0.5 - 2 kev luminosity of @xmath42 is driven apparently by the feedback from evolved stars in form of stellar mass loss and ( type ) ia sne , because there is no evidence for an agn or recent massive star formation in the bulge .
the bipolar morphology of the truly diffuse soft x - ray emission further indicates that the outflow is probably influenced by the presence of strong vertical magnetic field , same as that observed in the central region of our galaxy .
the spectrum of the emission can be characterized by an optically - thin thermal plasma with a temperature of @xmath43 k @xcite .
similar analyses have been carried out so far only for a couple of other early - type galaxies ( ngc 3379 , @xcite ; ngc 5866 , @xcite ) .
the faint stellar contribution , which has a significantly different spectral shape from lmxbs , still needs to be carefully accounted for in the analysis of other galaxies . in any case , it is clear that the diffuse x - ray luminosity accounts for at most a few % of the energy input from ia sne alone in a normal early - type galaxy .
where does the missing feedback go ?
another remarkable puzzle about the diffuse hot gas in elliptical galaxies is the apparent low metal abundances .
as inferred from x - ray spectral fits , the abundances are typically sub - solar for low- and intermediate - mass galaxies with log(@xmath44 to about solar for more massive ones ( e.g. , @xcite ) , substantially less than what are expected from the ia sn enrichment ( see the discussion section ) .
furthermore , a number of galaxies show a significant iron abundance drop toward their central regions , where the stellar feedback should be the strongest . examples of this abundance drop include m87 @xcite , ngc 4472 , ngc 5846 @xcite , and ngc 5044 @xcite .
the drop and the low abundance , if intrinsic , would then indicate that only a small portion of metals produced by ia sne is observed ; the rest is either expelled or in a state that the present x - ray data are not sensitive to .
in summary , the following are the key discoveries made with observations of diffuse hot gas in and around normal galaxies : * the hot gas generally consists of two components : 1 ) the disk component , which is sensitive to the star formation rate per unit area ( e.g. , @xcite ) ; 2 ) the bulge component , the amount of which is proportional to the stellar mass .
significant diffuse soft x - ray emission is detected only within @xmath45 kpc of the disk and bulge of a galaxy . *
the gas typically has a mean characteristic temperature of @xmath46 , although there is evidence for a considerably high temperature component ( up to @xmath47 k ) in and around galactic disks with active star formation @xcite .
this high - temperature component may represent a significant galactic hot gaseous outflow .
the interaction of this outflow with cool gas ( via shocks , dynamic mixing , and/or charge exchanges ) is responsible for much of the low - temperature one ( e.g. , @xcite ) .
* the cooling rate , as inferred from the diffuse x - ray emission , is insignificant , compared to the expected stellar feedback alone .
the bulk of the feedback is missing , especially in inter - arm regions and in galactic spheroids .
* the hot cgm seems to have a very low density and may not account for the missing baryon matter expected in individual galactic dark matter halos . while the above results are generally consistent with those found for our own galaxy , the studies of nearby galaxies also show intriguing sub - structures in the diffuse hot gas distributions ( e.g. , the apparent positive radial temperature and metal abundance gradients in elliptical galaxies ) .
as shown above , the diffuse soft x - ray emission observed in and around galaxies clearly traces the global hot gas that is heated by the stellar and possibly agn feedback . by comparing the x - ray observations directly with physical models and/or simulations of the feedback , we may further study its dynamics , which is so far hardly known .
the following discussion is concentrated on the modeling of the feedback in galactic spheroids for their relative simplicity .
the presence of substantial cool gas and star formation , as in typical disk galaxies , would certainly add complications .
the missing feedback problem is particularly acute in so - called low @xmath48 spheroid - dominated galaxies .
empirically , the ia sn rate is @xmath49 $ ] , where @xmath50 is the k - band luminosity of a galaxy @xcite . adopting the color index @xmath51 for a typical spheroid
, we estimate the total mechanical energy injection from ia sne ( assuming @xmath52 each ) is @xmath53 $ ] .
the mechanical energy release of an agn can also be estimated empirically @xcite as @xmath54 ^ 2.$ ] therefore , averaged over the time , ia sne are energetically more important than the agn in a galactic spheroid with @xmath55 .
furthermore , sn blastwaves provide a natural distributed heating mechanism for hot gas in galactic spheroids @xcite .
additional continuous heating is expected from converting the kinetic energy of stellar mass loss from randomly moving evolved stars to the thermal energy . on the other hand
, the agn feedback , likely occurring in bursts with certain preferential directions ( e.g. , in form of jets ) , can occasionally result in significant disturbances in global hot gas distributions , as reflected by the asymmetric x - ray morphologies observed in some elliptical galaxies ( e.g. , @xcite ) .
it is not yet clear as to what fraction of the agn feedback energy is converted into the heating of the hot gas observed . in general , the lack of distributed cool gas in galactic spheroids makes it difficult to convert and release the thermal energy into radiation in wavelength bands other than the x - ray .
not only the mechanical energy , the gas mass from the stars and ia sne , especially heavy elements , is missing as well . the mass injection rate is expected to be @xmath56 m_\odot { \rm~yr^{-1}}$ ] with a mean iron abundance @xmath57@xmath58@xmath59 , where @xmath60 is the iron abundance of the stars while @xmath61 is the iron mass yield per ia sn ( e.g. , @xcite ) .
the above empirical estimates of the energy and mass feedback rates , which should be accurate within a factor of @xmath21 , are typically a factor of @xmath62 greater than what are inferred from the diffuse x - ray emission .
naturally , one would expect that the missing feedback is gone with a wind ( outflow ) , spheroid - wide or even galaxy - wide .
the notion that ia sne may drive galactic winds is not new ( e.g. , @xcite ) . but could such winds explain the diffuse x - ray emission of galactic spheroids ? it is easy to construct a 1-d steady - state supersonic wind model , assuming that the specific energy of the feedback ( per mass ) of a galaxy is large enough to overcome its gravitational bounding and that the cgm pressure ( thermal or ram ) is negligible ( e.g. , @xcite ) .
this supersonic wind model depends primarily on two feedback parameters : the integrated energy and mass input rates
. however , the model in general fails miserably : it predicts a too low luminosity ( by a factor of @xmath62 ) , a too high temperature ( a factor of a few ) , and a too steep radial intensity profile to be consistent with observations of low @xmath63 galactic spheroids , in particular those in m31 and m104 @xcite .
only few very low - mass and gas - poor spheroids , hence very faint in x - ray emission , still seem to be consistent with the 1-d wind model ( e.g. , @xcite ) .
, logarithmically . *
( b ) * simulated 2-d large - scale gas density distribution in the @xmath64 plane ( in units of kpc ) . a nearly vertical magnetic field , similar to what is observed in the inter - cloud medium of the galactic center , is included in the simulation to test its confinement effect on the spheroid wind ; a reverse - shock is clearly visible . also apparent are instabilities at the contact discontinuity between the shocked spheroid wind gas and the ejected materials from the initial starburst .
the density is scaled in units of m@xmath65 . , scaledwidth=90.0% ] to compare with the x - ray emission , in fact one needs to account for 3-d effects .
x - ray emission is proportional to the emission measure and is thus sensitive to the detailed structure of hot gas in a galactic spheroid . to realistically generate the inhomogeneity in the heating and chemical enrichment processes , tang & wang have developed a scheme to embed adaptively selected 1-d snr seeds in 3-d spheroid - wide simulations of supersonic winds or subsonic outflows ( e.g. , @xcite ) .
these 3-d simulations , reaching a resolution down to adaptively refined scales of a few pc ( e.g. , fig . [
fig : f4]a ) , show several important 3-d effects @xcite : * soft x - ray emission arises primarily from relatively low temperature and low abundance gas shells associated with sn blastwaves .
the inhomogeneity in the gas density and temperature substantially alters the spectral shape and leads to artificially lower metal abundances ( by a factor of a few ) in a spectral fit with a simplistic thermal plasma model .
* reverse shock - heated sn ejecta , driven by its large buoyancy , quickly reaches a substantially higher outward velocity than the ambient medium .
the ejecta is gradually and dynamically mixed with the medium at large galactic radii and is also slowly diluted and cooled by _
insitu _ mass injection from evolved stars .
these processes together naturally result in the observed positive gradient in the average radial iron abundance distribution of the hot gas , even if mass - weighted .
* the average 1-d and 3-d simulations give substantially different radial temperature profiles ; the inner temperature gradient in the 3-d simulation is positive , mimicking a `` cooling flow '' .
( thus the study of nearby galactic spheroids may provide important insights into the behavior of the intragroup or intracluster gas around central galaxies , for which sporadic agn energy injections somewhat mimic the sn heating considered here . ) . *
the inhomogeneity also enhances the diffuse x - ray luminosity by a factor of a few in the supersonic wind case and more in a subsonic outflow .
in addition , a subsonic outflow can have a rather flat radial x - ray intensity distribution . * the dynamics and hence the luminosity of a subsonic flow
further depend on the spheroid star formation and feedback history .
this dependence , or the resultant luminosity variance , may then explain the observed large dispersion in the x - ray to k - band luminosity ratios of elliptical galaxies .
all considered , subsonic outflows appear to be most consistent with the x - ray observations of diffuse hot gas in typical intermediate - mass galactic spheroids and elliptical galaxies . whether a galactic outflow is supersonic or subsonic depends not only on the specific energy of the ongoing feedback , but also on the properties of the cgm , which is a result of the past interplay between the feedback and the accretion of a galaxy or a group of galaxies from the intergalactic medium ( igm ) .
tang et al . have illustrated how this interplay may work , based on several 1-d hydrodynamic simulations in the context of galaxy formation and evolution @xcite .
they approximate the feedback history as having two distinct phases : ( 1 ) an early starburst during the spheroid formation ( e.g. , as a result of rapid galaxy mergers ) and ( 2 ) a subsequent long - lasting and slowly declining injection of mass and energy from evolved low - mass stars .
an energetic outward blastwave is initiated by the starburst ( including the quasar / agn phase ) and is sustained by the long - lasting stellar feedback .
even for a small galactic spheroid such as the one in the milky way , this blastwave may heat up the cgm on scales beyond the present virial radius , thus the gas accretion from the igm into the galactic halo could be largely reduced ( see also @xcite for similar results from 3-d cosmological structure formation simulations ) .
the long - lasting stellar feedback initially drives a galactic spheroid wind ( fig .
[ fig : f4]b ) .
as the mass and energy injection decreases with time , the feedback may evolve into a subsonic and quasi - stable outflow .
this feedback / cgm interplay scenario provides a natural explanation to various observed phenomena : * it solves the missing feedback problem as discussed earlier ; the energy is consumed in maintaining the hot cgm and preventing it from fast cooling . *
the very low density of the cgm explains the dearth of the chemically enriched hot gas observed around galaxies ; much of the cgm or the intragroup gas has been pushed away to larger scales ( see also @xcite ) . *
the predicted high temperature is consistent with observations of the large - scale hot intragroup medium , which seems always higher than @xmath66 k @xcite . *
the cooling of the material ejected early ( e.g. , during the initial starbursts ; @xcite ) provides a natural mechanism for high - velocity clouds with moderate metal abundances ( fig .
[ fig : f4]b ) . such clouds may be seen in 21 cm line emission and in various absorption lines ( e.g. , ly@xmath18 , mg ii , si iii ; @xcite ) .
o vi line absorptions can then arise from either photo - ionization of the clouds or collisional ionization at their interfaces with the pervasive hot cgm .
based on complementary high - resolution imaging and spectroscopic observations from , we have learned a great deal about various high - energy phenomena and processes in galaxies : * the luminosity functions and their dependence on the stellar mass and star formation rate are nearly universal for essentially all major types of discrete x - ray sources in galaxies . *
a very hot component of the global ism in the galaxy has been revealed , and its global spatial , thermal , chemical , and kinetic properties have been characterized . * stringent upper limits
have been obtained to the contents of the chemically - enriched cgm around our galaxy and around other galaxies / groups along the sight lines toward several luminous agns . *
detailed structures of the superwinds emanated from starburst galaxies are resolved , tracing strong interplay between the very hot outflowing gas and the presence of cool gas . *
sporadic energy injections from agns are shown to play an important role in shaping the global hot gas in and around elliptical galaxies . *
truly diffuse hot gas has also been mapped out for a few nearby normal galaxies , indicating that ongoing stellar feedback may play an important role in regulating the galactic eco - systems . in conclusion , the existing work has demonstrated the power of observations in probing the stellar and agn feedback and its effect on galaxy evolution as well as inventorying various kinds of high - energy sources in galaxies .
i thank the referee for constructive comments and the organizers of the chandra s first decade of discovery conference for the invitation to give the talk that this paper is based on and am grateful to my students and collaborators for their contributions to the work described above , particularly yangsen yao who helped to produce figures 1 and 2 . the work is partly supported by nasa / cxc under grants g08 - 9088b and g08 - 9047a .
futamoto k , kazuhisa m , yoh t , ryuichi f , and noriko y y ( 2004 ) _ detection of highly ionized o and ne absorption lines in the x - ray spectrum of 4u 1820 - 303 in the globular cluster ngc 6624 _ , apj , 605 , 793 - 799 .
yao y , and wang q d ( 2006 ) _ x - ray absorption spectroscopy of the multiphase interstellar medium : oxygen and neon abundances _ , apj , 641 , 930 - 937 .
yao y , and wang q d ( 2007 ) _ the nonisothermality and extent of galactic diffuse hot gas toward markarian 421 _ , apj , 658 , 1088 - 1095 .
humphrey p j , buote d a , brighenti f , gebhardt k , and mathews w g ( 2009 ) _ hydrostatic gas constraints on supermassive black hole masses : implications for hydrostatic equilibrium and dynamical modeling in a sample of early - type galaxies _ , apj , 703 , 1257 - 1277 .
strickland d k , heckman t m , colbert e j m , hoopes c g , and weaver k a ( 2004 ) _ a high spatial resolution x - ray and h@xmath18 study of hot gas in the halos of star - forming disk galaxies .
i. spatial and spectral properties of the diffuse x - ray emission _
, apjs , 151 , 193 - 236 .
strickland d k , heckman t m , colbert e j m , hoopes c g , and weaver k a ( 2004 ) _ a high spatial resolution x - ray and h@xmath18 study of hot gas in the halos of star - forming disk galaxies .
ii . quantifying supernova feedback _ , apj , 606 , 829 - 852 .
tllmann r , pietsch w , rossa j , breitschwerdt d , and dettmar r j ( 2006 ) _ the multi - phase gaseous halos of star - forming late - type galaxies . i. xmm - newton observations of the hot ionized medium _ , a&a , 448 , 43 - 75 tllmann r , breitschwerdt d , rossa j , pietsch w , and dettmar r j ( 2006 ) _ the multi - phase gaseous halos of star - forming late - type galaxies .
ii . statistical analysis of key parameters _ ,
a&a , 457 , 779 - 785 .
li z , and wang q d ( 2007 ) _ chandra detection of diffuse hot gas in and around the m31 bulge _ , apjl , 668 , 39 - 42 .
david l p , jones c , forman w , vargas i m , and nulsen p ( 2006 ) _ the hot gas content of low - luminosity early - type galaxies and the implications regarding supernova heating and active galactic nucleus feedback _ , apj , 653 , 207 - 221 .
shull j m , jones j r , danforth c w , and collins j a ( 2009 ) _ a large reservoir of ionized gas in the galactic halo : ionized silicon in high - velocity and intermediate - velocity clouds _ , apj , 699 , 754 - 767 | this presentation reviews chandra s major contribution to the understanding of nearby galaxies . after a brief summary on significant advances in characterizing various types of discrete x - ray sources , the presentation focuses on the global hot gas in and around galaxies , especially normal ones like our own .
the hot gas is a product of stellar and agn feedback the least understood part in theories of galaxy formation and evolution .
chandra observations have led to the first characterization of the spatial , thermal , chemical , and kinetic properties of the gas in our galaxy .
the gas is concentrated around the galactic bulge and disk on scales of a few kpc .
the column density of chemically - enriched hot gas on larger scales is at least an order magnitude smaller , indicating that it may not account for the bulk of the missing baryon matter predicted for the galactic halo according to the standard cosmology .
similar results have also been obtained for other nearby galaxies .
the x - ray emission from hot gas is well correlated with the star formation rate and stellar mass , indicating that the heating is primarily due to the stellar feedback . however
, the observed x - ray luminosity of the gas is typically less than a few percent of the feedback energy .
thus the bulk of the feedback ( including injected heavy elements ) is likely lost in galaxy - wide outflows .
the results are compared with simulations of the feedback to infer its dynamics and interplay with the circum - galactic medium , hence the evolution of galaxies . |
the coalescence of neutron star - neutron star ( ns - ns ) and black hole - neutron star ( bh - ns ) binaries is one of the most promising sources of gravitational waves to be detected by ground based gravitational wave detectors like virgo and ligo @xcite .
these detectors have now reached the planned sensitivity and they will evolve toward a second generation , the advanced ( virgo and ligo ) detectors , with a sensitivity enhanced by an order of magnitude .
furthermore , a design study for an even more sensitive third generation detector , et ( einstein telescope ) , is in progress @xcite . in a recent study , based on a population synthesis approach @xcite , the formation and
the evolution of compact binary systems has been followed from the onset of star formation , both in the galactic field , where their massive binary progenitors evolve in isolation , and in dense clusters , where they can form at high rates due to dynamical interactions .
the authors estimate that advanced ligo should detect the merger of ns - ns binaries at a rate of @xmath1 events per year and the merger of bh - ns binaries at a rate of @xmath2 per year .
similar estimates hold for the advanced version of virgo .
thus , it is reasonable to expect that in a near future we may be able to detect the gravitational wave ( gw ) signals emitted by these sources and study their features .
a further reason to be interested in these coalescing binaries is that they have been proposed as providing the engine for short gamma - ray bursts ( sgrbs ) ; the detection of a gravitational wave signal emitted by one of these systems in coincidence with a sgrb would validate this model , thus clarifying one of the most interesting open issues in astrophysics . in this paper
we study the disruption of a neutron star in a bh - ns coalescing binary , to envisage a method to extract information on the equation of state ( eos ) of matter in the ns interior from the detection of the gravitational wave signal emitted in the process .
we use the _ affine model _
approach @xcite,@xcite,@xcite,@xcite , which treats the ns as an extended object , which responds to its self - gravity , to its internal pressure forces and to the relativistic bh tidal field , under the constraint that its shape is always that of an ellipsoid . in the original formulation of this model , the internal structure of the star was treated at a newtonian level and the eos was assumed to be polytropic . in @xcite
( to be referred to as paper i hereafter ) we have improved this approach by introducing a general relativistic description of the stellar structure , and generalizing the relevant equations to include more general equations of state .
this improved approach has been applied to study quasi - equilibrium sequences of bh - ns binaries , and to determine the critical orbital separation at which the star is torn apart by the black hole tidal field .
when the ns is disrupted before reaching the innermost stable circular orbit ( isco ) , the emitted gw signal is expected to change abruptly and its amplitude is expected to decrease sharply ; such signal should exhibit a cutoff frequency @xmath0 , corresponding to the orbit at which the disruption occurs , i.e. @xmath3 . this prediction
is confirmed by numerical simulations where the coalescence of a black hole and a neutron star and the tidal disruption of the star have been studied i ) in the framework of newtonian gravity @xcite , @xcite , and ii ) in full general relativity @xcite , @xcite .
the frequency cutoff is a distinctive feature of the waveform emitted by a bh - ns coalescing binary and indicates the disruption of the star .
this was first pointed out in @xcite , where a relation between @xmath0 and the stellar radius was derived , using the formulation of the affine model appearing in @xcite and describing the inspiralling of an incompressible , homogeneous , newtonian ellipsoid moving on circular orbits around a rotating black hole .
hints on the role of the compressibility were derived in @xcite using some results of @xcite , where the same process was studied for a polytropic , newtonian star .
in this paper we study the relation between the cutoff frequency and the neutron star eos . using the same approach developed in paper i
, we explicitly compute for which values of the masses of the binary components and of the black hole angular momentum the star is tidally disrupted before reaching the isco , for a variety of realistic equations of state .
knowing the disruption distance @xmath4 , we evaluate @xmath0 and show how a measure of this quantity from a detected gw signal may be used to infer interesting information on the equation of state of matter in the neutron star interior .
it is known that the fully relativistic numerical study of the last phases of bh - ns binary coalescence is a quite difficult task , especially for large values of the mass ratio @xmath5 ; this is due to a number of reasons , which include the lack of symmetry , the difficulty in evolving the ns while handling the bh singularity , and the prohibitively high computational costs required to span the parameter space ( mass ratio , eos , spins etc . ) . for these reasons ,
the literature on the subject is limited to a restricted number of studies , most of which are in the framework of newtonian gravity .
bh - ns coalescence in general relativity has been studied in the following papers ( see @xcite for a recent review on the subject ) : non spinning black hole and @xmath6 in @xcite ; @xmath7 in @xcite ; @xmath8 in@xcite ; and @xmath9 , @xmath10 and @xmath11 in @xcite . in @xcite ,
quasi - equilibrium sequences of black hole - neutron star binaries have been studied in general relativity for @xmath12 .
the case of a rotating black hole has been considered in @xcite , where the coalescence has been studied for three values of the black hole angular momentum ( @xmath13 ) and for @xmath14 , whereas in @xcite the case @xmath15 has been investigated .
it should be stressed that in _ all _ these studies the neutron star is modeled using a polytropic equation of state .
the improved affine model approach which we use in our study is an approximate method , but it has the advantage of allowing one to explore a large region of the parameter space , including large values of the mass ratio @xmath5 , and to use modern equations of state to model matter in the neutron star interior , at a much lower computational cost than fully relativistic simulations . with respect to ref .
@xcite our study introduces several novelties : we describe the internal structure of the star using general relativity , we explicitly take into account the stellar compressibility and describe the matter inside the star using realistic equations of state ; in addition , in @xcite @xmath0 was provided by the kerr geometry in the approximation of a point mass ns following geodesics in this geometry , while we determine this quantity using the 2.5 post newtonian equations of motion for the orbital dynamics of a binary system .
the plan of the paper is the following . in section [ affinemodel ]
we briefly describe the affine approach ; in section [ nutide ] we discuss how to determine the cutoff frequency @xmath0 .
the equations of state we employ are briefly described in section [ eos ] ; in section [ errorbar ] we discuss the errors which affect the measure of the binary parameters from a detected gw signal , and how they propagate and affect the evaluation of @xmath0 and of the neutron star radius .
the results of our study are reported in section [ results ] and in section [ concl ] we draw our conclusions .
the equations of the improved affine model approach , which we need in order to determine the radial distance @xmath4 at which the neutron star is torn apart by the tidal interaction with the black hole , are described in detail in paper i. here we will summarise the assumptions underlying this approach and we shall write only the equations to be solved .
we consider a star in the tidal field of a kerr bh whose center of mass follows equatorial , circular orbits ; while moving , the star maintains an ellipsoidal shape ; more precisely it is a riemann - s type ellipsoid , i.e. its spin and vorticity are parallel and their ratio is constant ( see @xcite ) .
the ns equilibrium structure is determined using the stellar structure equations of general relativity , while dynamical behaviour is governed by newtonian hydrodynamics improved by the use of an effective relativistic self - gravity potential .
the equations for the ns deformations are written in the principal frame , i.e. the frame associated with the principal axes of the stellar ellipsoid .
tidal effects on the orbital motion are neglected , as well as the perturbation that the star induces on the bh .
we study the evolution of the system in the quasi - equilibrium approximation , i.e. we neglect all time derivatives appearing in the ordinary differential equations of the model .
physically this means that we assume that the ns follows a quasi - equilibrium sequence during the bh - ns coalescence ; according to the model , moreover , the circulation @xmath16 of the fluid along this sequence is constant . in particular , we set @xmath17 , that is , we consider the ns fluid to be irrotational . in the principal frame ,
the fluid variables of the improved affine model are five : the three principal axes of the stellar ellipsoid @xmath18 and @xmath19 , the angular frequency of the internal fluid motion @xmath20 and the star spin @xmath21 .
the axes @xmath22 and @xmath23 both belong to the orbital plane , while @xmath19 is perpendicular to it ; @xmath22 indicates the axis that lies along the binary orbital radius .
the quasi - equilibrium equations governing these variables are : @xmath24 where @xmath25 and @xmath26 are , respectively , the effective relativistic self - gravity potential and the scalar quadrupole moment of the ns at spherical equilibrium ( see below ) .
@xmath27 is the ns radius and the @xmath28 s are defined as @xmath29 the @xmath30 s denote the components of the bh tidal tensor in the principal frame ; @xmath31 which is connected to @xmath21 by @xmath32 is the angle that brings the parallel - propagated frame in the principal frame , by a rotation around the @xmath19 axis @xcite .
finally , @xmath33 is an angle that governs the rotation of the parallel - propagated frame in order to preserve its parallel transport , and its time evolution is given by @xmath34 where @xmath35 and @xmath36 are the bh mass and the bh - ns orbital separation respectively .
notice that the last equation and eq.([eq : phi = psi ] ) imply that @xmath37 the fifth equation for the five fluid variables is provided by the definition of the circulation @xmath38\ , , \nonumber\\\end{aligned}\ ] ] so that for irrotational fluids one has @xmath39 the effective relativistic self - gravity potential is given by @xmath40 where @xmath41 is given by the tolman - oppenheimer - volkoff ( tov ) stellar structure equations @xmath42[m_{tov}(r ) + 4\pi r^3p(r)]}{\rho(r ) r[r-2m_{tov}(r)]}\nonumber\\ \label{def : mtov } m_{tov}(r ) & = & 4\pi\int_0^r dr'r'^2\epsilon(r')\,.\nonumber\end{aligned}\ ] ] the scalar quadrupole moment @xmath26 is defined as @xmath43 and , like @xmath25 , must be calculated at spherical equilibrium . finally , the relevant kerr bh tidal tensor components are @xmath44\\ c_{22 } & = & \frac{m_{bh}}{r^3}\left[1 - 3\frac{r^2+k}{r^2}\sin^2(\psi-\phi ) \right]\\ \label{def : c33 } c_{33 } & = & \frac{m_{bh}}{r^3}\left ( 1 + 3\frac{k}{r^2}\right)~,\end{aligned}\ ] ] where @xmath45 @xmath46 is the black hole spin parameter , and @xmath47 and @xmath48 are , respectively , the energy and the @xmath49-orbital angular momentum per unit mass . since we consider circular equatorial geodesics , @xmath47 and @xmath48 are @xmath50 where @xmath51 we solve eqs .
( [ eq : a1])-([eq : a3 ] ) , with the aid of eqs.([eq : omega ] ) , ( [ eq : lambda ] ) and of definitions ( [ def : tildea ] ) , ( [ def : whv ] ) , ( [ def : whm ] ) , ( [ def : c11])-([def : c33 ] ) , by adopting a multidimensional newton - raphson scheme @xcite in order to determine the values of the axes of the ellipsoid for each quasi - stationary orbit , identified by the orbital separation @xmath36 . we start by solving the tov stellar structure equations for a non - rotating spherical neutron star in equilibrium .
we then fix the black hole spin parameter @xmath46 and the binary mass ratio @xmath52 and place the star at a distance @xmath53 from the black hole ( we obviously make sure that the sequence we obtain is independent of @xmath54 ) .
subsequently , we gradually reduce the orbital separation and solve eqs.([eq : a1])-([eq : a3 ] ) at each step and monitor the star axes until a critical separation @xmath4 is reached , at which a quasi - equilibrium configuration is no longer possible .
this critical distance physically corresponds to the tidal disruption of the neutron star .
it may be identified by exploiting the fact that the newton - raphson algorithm can not find any solution to the system of equations , or by calculating numerically the derivative @xmath55 , where @xmath56 and keeping track of it since it tends to zero at tidal disruption : both methods yield the same values of @xmath4 .
we mention that an alternative approach to evaluate @xmath4 in a newtonian framework , based on the estimate of the roche lobe radius , has been used in @xcite .
if the tidal disruption is not encountered , the quasi - equilibrium ends when the neutron star surface crosses the black hole horizon and hence the coalescence terminates with a plunge . the quantity @xmath4 has to be compared with the value of the radius of the innermost stable circular orbit @xmath57 , which is determined by using the formulae derived in @xcite for a point mass in the gravitational field of a kerr bh : @xmath58^{1/2}\}\nonumber\\ z_1 & = & 1+(1-a^2/m_{bh}^2)^{1/3}\nonumber\\ & \times&[(1+a / m_{bh})^{1/3}+(1-a / m_{bh})^{1/3}]\nonumber\\ z_2 & = & ( 3a^2/m_{bh}^2+z_1 ^ 2)^{1/2}\ , , \label{risco}\end{aligned}\ ] ] where the upper ( lower ) sign holds for co- ( counter-)rotating orbits .
we remind the reader that if @xmath59 the star is disrupted before the merger starts , and the gravitational signal will exhibit a cutoff at a frequency @xmath0 .
to compute @xmath0 , we model the inspiral of the mixed binary by means of a post - newtonian ( pn ) approach and truncate the inspiral when the orbital separation reaches @xmath4 ; we then read off the orbital frequency at the tidal disruption @xmath60 : this is related to the gw frequency by @xmath61 .
we follow a hamiltonian approach .
the conservative part of the two - body hamiltonian is known up to order @xmath62pn , e.g. @xcite ; however , we shall truncate it at @xmath63pn order since we will use gw dissipation terms of order @xmath64pn @xcite . the pn - expanded hamiltonian for the relative motion of the bh - ns binary is @xmath65 the single contributions being : @xmath66\nonumber\\ & + & \frac{g_{n}^2m_{tot}^2}{2c^4\mu r^2 } \left[(5
+ 8\eta)\left ( p_r^2 + \frac{p_\varphi^2}{r^2}\right)+3\eta p_r^2\right ] \nonumber\\ & -&\frac{g_{n}^3(1 + 3\eta)\mu m_{tot}^3}{4c^4r^3}\\ \label{eq : h_so } \mathcal{h}_{so } & = & \frac{g_{n}}{c^2r^3}{\mathbf l } \cdot\left(2+\frac{3m_{ns}}{2m_{bh}}\right){\mathbf j}_{bh}\\ \label{eq : h_ss } \mathcal{h}_{ss } & = & \frac{g_{n}}{c^2r^3 } [ 3({\mathbf j}_{bh}\cdot{\mathbf n})({\mathbf j}_{bh}\cdot{\mathbf n})\nonumber\\ & -&({\mathbf j}_{bh}\cdot{\mathbf j}_{bh } ) ] \frac{m_{ns}}{m_{bh}}\end{aligned}\ ] ] where @xmath67 is the gravitational constant , @xmath68 is the speed of light , @xmath69 is the total mass of the system , @xmath70 is its reduced mass , @xmath71 is the symmetric mass ratio , @xmath72 and @xmath73 are the conjugate variables of the orbital separation @xmath36 and the orbit angle coordinate @xmath74 , @xmath75 is the orbital angular momentum and @xmath76 is the spin angular momentum of the black hole .
the vectors @xmath77 and @xmath76 are both perpendicular to the orbit plane ; the magnitude of the latter is equal to @xmath78 and such vector is not evolved as a dynamic variable in order to follow the spirit of the affine model , according to which the presence of the ns does not influence the bh . the spin - orbit ( so ) and spin - spin ( ss ) contributions to @xmath79 therefore reduce to @xmath80 and @xmath81 the dynamics described by this hamiltonian is supplied with gw dissipation by means of the @xmath64pn non - conservative terms @xmath82 to be included in the hamilton equations of motion as @xmath83 within this hamiltonian approach , once the binary orbit is evolved until @xmath84 , the gw cutoff frequency due to the ns tidal disruption @xmath0 is given by @xmath85
the equation of state of matter at densities larger than the saturation density of nuclear matter , @xmath86 g/@xmath87 , is uncertain . at these densities , which are typical of a neutron star core
, neutrons can no longer be considered as non interacting particles , and different ways of modeling their interactions lead to a different composition , which may include heavy baryons , quarks etc .. all available eoss of strongly interacting matter have been obtained within models based on the theoretical knowledge of the underlying dynamics and constrained , as much as possible , by empirical data .
modern eoss are derived within two main , different approaches : non - relativistic nuclear many - body theory ( nmbt ) and relativistic mean field theory ( rmft ) .
as representative of nmbt we choose two eoss named apr2 and bbs1 . for apr2 matter
consists of neutrons , protons , electrons and muons ( @xmath88 ) in weak equilibrium .
the hamiltonian includes two- and three- nucleon interaction terms ; the two - nucleon term is the argonne @xmath89 potential @xcite , the three - nucleon term is the urbana ix potential @xcite .
the many - body schrdinger equation is solved using a variational approach @xcite .
the calculations include relativistic corrections to the two - nucleon potential , arising from the boost to a frame in which the total momentum of the interacting pair is non - vanishing .
these corrections are necessary to use phenomenological potentials , describing interactions between nucleons in their center of mass frame , in a locally inertial frame associated with the star .
the maximum mass for this eos is @xmath90 .
the matter composition of bbs1 is the same as in the apr2 model .
the eos is obtained using a slightly different hamiltonian , including the argonne @xmath89 two - nucleon potential and the urbana vii three - nucleon potential @xcite .
the ground state energy is calculated using g - matrix perturbation theory @xcite .
the maximum mass is @xmath91 . as representative of the rmft approach ,
we choose the eoss named bgn1h1 , gnh3 and bpal .
the balberg - gal ( bgn1h1 ) eos @xcite describes matter consisting of neutrons , protons , electrons , muons and hyperons ( @xmath92 , @xmath20 and @xmath93 ) in equilibrium .
the effective potential parameters are tuned in order to reproduce the properties of nuclei and hypernuclei according to high energy experiments .
this eos is a generalization of the lattimer - swesty eos @xcite , which does not include hyperons .
the maximum mass is @xmath94 .
the glendenning ( gnh3 ) eos @xcite considers @xmath88 up to a certain density @xmath95 ; beyond this point , additional baryon states ( such as the @xmath96 and the hyperons @xmath97 ) and the mesons @xmath98 are introduced . below the hadronization density @xmath99
the eos is very stiff but causal ; the appearance of hyperons strongly softens the eos because they are more massive than nucleons and when they start to fill their fermi sea they are slow and replace the highest energy nucleons .
the maximum mass is @xmath100 .
the three eoss bpal12 , bpal22 and bpal32 , are derived using a density dependent nucleon - nucleon effective interaction ( as for skyrme nuclear interactions ) @xcite .
matter is composed of @xmath88 in weak equilibrium , no hyperons are present and the eos parameters are fixed to reproduce the saturation properties of nuclear matter .
bpal12 is particularly soft , with a nuclear incompressibility @xmath101 mev .
since the empirical value commonly accepted ranges within @xmath102 mev , it is clear that bpal12 has to be considered as an eos soft extreme , but still compatible with astrophysical observations , since the maximum mass it predicts is @xmath103 .
bpal22 and bpal32 are two different versions of the bpal eos , corresponding to more realistic values of @xmath104 : @xmath105 mev for bpal22 , with a maximum mass @xmath106 , and @xmath107 mev for bpal32 , with a maximum mass @xmath108 . at a density of about one half of the nuclear saturation density
we match the eos of the core with a crust , which is composed of three layers ; the inner layer is the douchin - haensel ( sly4 ) crust @xcite ; for @xmath109g/@xmath87 and for @xmath110g/@xmath87 , we use , respectively the haensel - pichon ( hp94 ) eos @xcite , and the baym - pethick - sutherland ( bps ) eos @xcite . in fig .
[ fig1 ] we show the mass - radius diagram for the selected eoss .
they clearly exhibit very different behaviours . to some extent
, bpal12 and gnh3 can be considered , respectively , as soft and hard extremes .
apr2 has a wide mass interval where the radius is almost insensitive to mass variations ; bgn1h1 also shows this feature , albeit for a more restricted mass interval , whereas the bpal and gnh3 eoss do not .
bgn1h1 , on the other hand , has a sudden softening which does not characterise the other three equations of state ; this softening is due to the appearance of hyperons in the core of the neutron star above a critical central density .
let us suppose that in a bh - ns coalescence the neutron star is disrupted before reaching the isco , and that the gravitational wave interferometers detect the emitted signal which , as discussed in previous sections , has the form of a chirp truncated at the frequency @xmath0 . with a suitable data analysis
, the values of the symmetric mass ratio @xmath111 and of the chirp mass @xmath112 can be extracted from the data with a certain error .
these errors have been evaluated for non - spinning compact binary sources , including up to 3.5pn terms , in @xcite,@xcite .
for instance , they are displayed in fig .
12 of @xcite , which shows the one - sigma fractional errors in @xmath113 and @xmath114 for non - spinning binary black hole sources as a function of the total mass of the system .
these errors reduce when the dynamical evolution of spins is included , so that they may be regarded as upper limits .
since a similar analysis has not been performed in the bh - ns case , in what follows we shall adopt these errors as appropriate also for mixed binaries . in the following
we shall consider neutron stars whose mass is in the range @xmath115 solar masses and mass ratios from 3 to 15 , therefore the total mass will range within @xmath116 . in this region the fractional error on @xmath113 is smaller than @xmath117 and of the order of @xmath118 for @xmath114 .
these data refer to advanced ligo assuming the source at a fixed distance of @xmath119 mpc . since in our case the mass parameters of the binary are the neutron star mass and the mass ratio , we express @xmath113 and @xmath114 as = , = .
[ etamchirp ] by propagating the errors we find the following expressions for the one - sigma fractional errors in the neutron star mass and in the mass ratio : [ eq : qerr ] & = & + [ eq : mnserr ] & = & + .
the absolute error on @xmath120 and @xmath52 for the binaries we consider are given in table [ table1 ] .
._absolute errors on the neutron star mass and on the mass ratio .
@xmath121 is independent of the neutron star mass .
@xmath122 is expressed in solar mass units . _ [ cols="^,^,^,^,^",options="header " , ] [ table1 ] we shall now discuss how the errors which affect the estimate of the binary parameters influence the evaluation of the neutron star radius and of @xmath0 . since we do not know the error on the black hole angular momentum , in what follows we shall assume that , with an accurate post - detection data analysis , @xmath123 could be measured with a @xmath124 accuracy . as an example , let us suppose that the neutron star mass and the mass ratio measured from a detected signal are , say , @xmath125 and @xmath126 , respectively , plus or minus the corresponding error which can be found in table [ table1 ] . in fig .
[ fig2 ] we plot the neutron star radius versus @xmath0 , evaluated as explained in section [ nutide ] , for any possible combination of the following data : m_ns=(1.40.02 ) m _ , = 0.50.05 , q=50.15 , [ errors ] and for a given eos , for instance bpal12 .
the data are spread in a region which can be identified by a vertical and a horizontal error bar , indicated as a red cross in the figure : the vertical error corresponds to the maximum and minimum values of the neutron star mass .
the horizontal error is identified by the points @xmath127 and @xmath128 . in the following figures , where we will show the graphs of @xmath27 versus @xmath0 for different values of the binary parameters and for different eoss
, we shall plot directly the cross , determined as in fig .
the results of our calculations are summarised in the figs .
[ fig3]-[fig9 ] . in fig .
[ fig3 ] we plot the ns radius , @xmath27 , versus the cutoff frequency , @xmath0 , for a black hole with angular momentum @xmath129 , and for a neutron star with mass @xmath130 in fig .
[ fig3](a ) , and @xmath131 in fig .
[ fig3](b ) .
the data are plotted for different values of the mass ratio @xmath132 ( we omit writing `` @xmath133 '' in the figures ) and for the considered eoss .
the values of @xmath122 and @xmath121 are given in table [ table1 ] .
for each value of @xmath52 , the continuous black lines are parabolic fits of the data corresponding to each eos .
[ fig3 ] shows that , for a given eos , as @xmath52 increases @xmath0 decreases .
moreover , for @xmath134 the data corresponding to the eoss bbs1 , bpal22 , apr2 , bpal12 are missing .
this behaviour is easily understood if we plot the radius at which disruption occurs , @xmath4 , versus @xmath0 for different values of @xmath52 .
for instance , in fig .
[ fig4 ] this plot is done for @xmath135 and @xmath136 for the eoss bgn1h1 and bpal12 . for comparison
, we also plot the value of @xmath57 , given by eqs .
( [ risco ] ) , versus the frequency @xmath137 of the gravitational signal emitted when a point mass of mass @xmath138 , orbiting around a kerr black hole of mass @xmath139 and angular momentum @xmath136 , reaches @xmath57 . for the bgn1h1 star ( upper curve )
we see that @xmath4 is larger than @xmath57 for all values of @xmath52 ; thus the gw signal emitted by these systems will exhibit a frequency cutoff at @xmath0
. moreover , as @xmath52 increases the value of @xmath4 increases , i.e. the star is disrupted at larger distances from the black hole . as
a consequence @xmath0 is a decreasing function of @xmath52 , as shown in fig .
[ fig3 ] for all eoss .
a similar behaviour is shown by the eos bpal12 ( middle curve ) up to @xmath15 . at that point
the curve @xmath4-versus-@xmath0 crosses the lower curve @xmath57-versus-@xmath137 , and for larger @xmath52 @xmath4 becomes smaller than @xmath57 ; this means that , for @xmath140 , the bpal12 star would be disrupted after reaching the isco and the emitted signal would not exhibit a frequency cutoff . for this reason the part of curve between the points corresponding to @xmath15 and @xmath134
is indicated as a dashed line .
going back to fig .
[ fig3 ] , the above discussion clarifies why for @xmath134 the data for the eoss bbs1 , bpal22 , apr2 and bpal12 are missing : the star merges with the black hole without being disrupted , and there is no frequency cutoff in the emitted gw signal . in fig .
[ fig3 ] we also notice that the vertical errors are always much smaller than the horizontal ones , except for the eos bpal12 , when @xmath141 .
the reason is that the vertical error is due to the error on the neutron star mass , which we recall is of the order of a few percent ; the curves corresponding to the different eoss in the mass - radius diagram of fig .
[ fig1 ] show that , for @xmath135 or @xmath142 , the stellar radius does not change significantly for such a small change in @xmath120 .
however , @xmath135 is close to the maximum mass of the bpal12 eos , and in its neighborhood the mass - radius curve is almost flat . for this eos
even a small change in the mass corresponds to a significant change in the radius .
[ fig3](b ) shows that , when @xmath143 , for the pairs of eoss ( bbs1,bpal22 ) and ( apr2,bpal12 ) the data nearly coincide .
the reason is that , as shown in fig .
[ fig1 ] , for @xmath142 the stars corresponding to these pairs of eoss have nearly the same radius . for the same reason
the values of @xmath0 for bgn1h1 and bpal32 are very close . in fig .
[ fig5 ] the same quantities of fig .
[ fig3 ] are plotted for a larger mass , @xmath144 .
the data for the eos bpal12 are missing because the maximum mass allowed by this eos is @xmath103 . a comparison with fig .
[ fig3](a ) shows that , for @xmath145 , the data for the eos bgn1h1 move at the bottom of the figure , i.e. for these masses the ns radius is smaller than that given by the other eoss , as can also be seen from fig .
[ fig1 ] ; moreover , the vertical error bars are larger , because for this eos @xmath145 is very close to the maximum mass , where the mass - radius curve is nearly flat ( see fig . [ fig1 ] ) . in fig .
[ fig6 ] the same quantities of fig .
[ fig3 ] are plotted for a black hole with angular momentum @xmath146 .
[ fig6](a ) refers to a neutron star with mass @xmath147 , whereas in fig .
[ fig6](b ) @xmath148 .
the data for @xmath15 and @xmath149 are missing because no disruption occurs for the considered eos : tidal disruption is favoured by large values of the black hole angular momentum .
if we compare this figure with fig .
[ fig3 ] we see that , for a given mass and eos , the value of the cutoff frequency is smaller if the black hole angular momentum decreases .
this means that the tidal disruption radius @xmath4 increases as @xmath46 decreases . it should be noted that for a fixed value of @xmath120 and of @xmath52 @xmath57 also increases as @xmath46 decreases ; it increases faster than @xmath4 , therefore there exists a critical value of the black hole angular momentum _
below _ which the star is not disrupted before the isco . in a similar way ,
for fixed @xmath46 , @xmath120 and eos , there exists a critical value of @xmath52 _ above _ which no disruption occurs .
eos & & + & @xmath150 & @xmath151 ( hz ) & @xmath150 & @xmath151 ( hz ) + gnh3 & @xmath152 & @xmath153 & @xmath154 & @xmath155 + bgn1h1 & @xmath156 & @xmath157 & @xmath158 & @xmath159 + bpal32 & @xmath160 & @xmath161 & @xmath162 & @xmath163 + bbs1 & @xmath164 & @xmath165 & @xmath166 & @xmath167 + bpal22 & @xmath164 & @xmath168 & @xmath169 & @xmath170 + apr2 & @xmath171 & @xmath172 & @xmath173 & @xmath174 + bpal12 & @xmath162 & @xmath175 & @xmath176 & @xmath177 + + gnh3 & @xmath178 & @xmath179 & @xmath180 & @xmath181 + bgn1h1 & @xmath182 & @xmath183 & @xmath184 & @xmath185 + bpal32 & @xmath186 & @xmath187 & @xmath188 & @xmath189 + bbs1 & @xmath190 & @xmath191 & @xmath192 & @xmath193 + bpal22 & @xmath194 & @xmath195 & @xmath196 & @xmath197 + apr2 & @xmath198 & @xmath199 & @xmath200 & @xmath201 + bpal12 & @xmath202 & @xmath203 & @xmath204 & @xmath205 + [ table2 ] as an example , in table [ table2 ] we give the values of @xmath150 and the corresponding @xmath0 , which is the minimum value of the cutoff frequency to be expected , for the binaries considered in figs . [ fig3 ] and [ fig6 ]
. a direct comparison of the effect of the angular momentum on @xmath0 is displayed in fig .
[ fig7 ] where the plot of @xmath27 versus @xmath0 is done for @xmath206 , @xmath207 and for @xmath208 and @xmath209 .
we see that the effect is small , but not negligible .
an interesting feature common to the figs .
[ fig3 ] , [ fig5 ] and [ fig6 ] is that , for the considered eoss , all data are well fitted by a parabolic fit .
it should be noted that the fit proposed by vallisneri in ref .
@xcite , i.e. \ { rr 0.145(m_bh)^-0.71 & m_bh 0.045 + 0.069(m_bh)^-0.95 & m_bh 0.045 + .
[ vallisn ] ( where @xmath210 ) for @xmath135 and @xmath211 , predicts values of @xmath27 and @xmath0 largely different from those we find , as shown in fig .
[ fig8 ] . in particular we see that
, for a given value of the ns mass , of the radius ( i.e. of the stellar compactness ) and of the mass ratio , the value of @xmath0 evaluated in @xcite is systematically smaller than the value we find .
this happens for a number of reasons .
for instance , let us consider two binary systems having a black hole of the same mass and a neutron star with the same mass and compactness ; in one case the ns structure is computed using the equations of newtonian gravity and an @xmath212 polytropic eos , in the other case using the tov equations and one of the eos we consider in this paper .
if we evaluate the value of @xmath4 for the first system using the affine approach in newtonian gravity as in @xcite , and for the second system using our improved affine approach , we will always find @xmath213 . as a consequence
the value of @xmath0 evaluated by the newtonian approach will be smaller than that evaluated with our improved approach .
furthermore , given the value of @xmath4 , in @xcite @xmath0 is calculated using the formula for a point mass moving on a circular orbit in kerr spacetime , while we compute this quantity using the 2.5 post newtonian equations describing the orbital evolution of a binary system .
this introduces a further difference which makes our @xmath0 larger than that evaluated with the geodesic approximation , and the difference increases when we consider small values of the mass ratio @xmath52 , as shown in fig .
[ fig8 ] .
the parameters @xmath214 of our parabolic fits r_ns=+_gwtide+_gwtide^2 , where @xmath27 is expressed in km and @xmath0 in hz , are given in table [ table3 ] .
& & + @xmath52 & @xmath215 & @xmath216 & @xmath217 & @xmath215 & @xmath216 & @xmath217 + 3 & 34.7&-2.8@xmath218&8.1@xmath219 & 32.4&-2.3@xmath218&5.5@xmath219 + 5&38.5&-3.6@xmath218&11.9@xmath219 & 35.3&-2.9@xmath218&7.5@xmath219 + + 3 & 33.2&-2.5@xmath218&6.9@xmath219 & 31.7&-2.1@xmath218&5.0@xmath219 + 5 & 37.1&-3.3@xmath218&10.2@xmath219 & 37.4&-3.2@xmath218&9.0@xmath219 + 10 & 50.0&-6.7@xmath218&27.7@xmath219 & 56.3&-8.0@xmath218&34.5@xmath219 + 15 & 47.9&-7.1@xmath218&32.4@xmath219 & 189.8&-0.47&31.4@xmath220 + [ table3 ] we shall now discuss how the results of this paper could be used to estimate the radius of the star and to gain information on the equation of state , using gravitational wave detection .
let us assume that a gravitational wave signal has been detected and exhibits a frequency cutoff at , say , @xmath221 hz .
a suitable data analysis on the chirp part of the signal allows to determine the neutron star mass , the mass ratio , with the corresponding uncertainty given in table [ table1 ] , and the black hole angular momentum with , say , an error of about 10% .
let us suppose that their values are @xmath222 , @xmath223 and @xmath129 , with @xmath122 and @xmath121 given in table [ table1 ] . with this information , we compute the ns radius versus @xmath0 for the considered eoss as explained in section [ compute ] , and plot the results in fig . [ fig9 ] .
the two continuous black lines are the parabolic fits of the points which , for each eos , are at the extreme of the horizontal error bar . if we draw a vertical line corresponding to the supposed detected cutoff frequency , i.e. @xmath221 hz , we see that it intercepts the two lines in two points , which identify a region within which the stellar radius should fall . the data shown in fig . [ fig9 ]
tell us that we would determine the ns radius with an error of 2.5% .
the error would be smaller if the detected value of @xmath0 is smaller , and larger if it is larger .
for instance for @xmath224 hz it would be 3.3% , and 2.2% for @xmath225 hz .
moreover , we would be able to exclude the eoss which fall outside the region framed by the horizontal dashed lines , putting strict constraints on the equation of state inside the neutron star .
in this paper we show that , by detecting a gravitational wave signal emitted by a bh - ns coalescing binary which exhibits a frequency cutoff due to the disruption of the star before the isco , we may be able to determine the radius of the star with quite a good accuracy , of the order of a few percent , and to put strict constraints on the equation of state of matter in the neutron star core .
our study does not intend to be exhaustive , since many more equations of state may be considered in the analysis ; for instance we did not consider quark stars .
there may exist other branches in the plots shown in figs .
[ fig3]-[fig7 ] corresponding to more exotic eoss .
however , from our study it emerges that the quantity which mostly affects @xmath0 is the stellar compactness ; therefore , we expect that in general the cutoff frequencies lay on the parabolic fits which correspond to a given neutron star mass , mass ratio and black hole angular momentum .
a further point which should be discussed is the following . in our analysis
we have assumed that the value of @xmath0 is known from the detection of a gravitational wave signal , but of course this quantity also is affected by uncertainties . for instance , we do not know how quickly the amplitude of the chirp goes to zero at tidal disruption , and therefore how sharp the step in the fourier transform of the gravitational signal @xmath226 at @xmath0 is . to answer this question , an accurate modeling of the bh - ns coalescence process
is certainly needed . moreover , assuming a given `` slope '' in @xmath226 at tidal disruption , how large would the experimental error in the determination of @xmath0 be ?
we plan to investigate this problem with a suitable data analysis study on the data of virgo .
we are indebted to ignazio bombaci for kindly providing the data for the eoss bpal12 , 22 and 32 , and to omar benhar for useful suggestions and discussions .
this work was partially supported by compstar , a research networking program of the european science foundation .
l.g . has been partially supported by grant no .
ptdc / fis/098025/2008 .
f.p . was supported in part by dfg grant sfb / transergio7 gravitational wave astronomy . . .
a. sadowski , k. belczynski , t. bulik , n. ivanova , f.a .
rasio and r. oshaughnessy , astrophys . j. * 676 * , 1162 , ( 2008 ) .
b. carter , j.p .
luminet , mon . not .
r. astron .
* 212 * , 23 , ( 1985 ) .
luminet , j.a .
marck , mon . not .
r. astron
* 212 * , 57 , ( 1985 ) .
p. wiggins , d. lai , astrophys .
j. * 532 * , 530 , ( 2000 ) .
lattimer , d.n .
schramm , astrophys .
j. * 210 * , 549 ( 1976 ) .
v. ferrari , l. gualtieri , f. pannarale , class .
* 26 * 125004 , ( 2009 )
. w. h. lee , w. kluzniak , mon . not .
r. astron
. soc . *
308 * , 780 , ( 1999 ) .
w. h. lee , mon . not .
r. astron
. soc . * 328 * , 583 , ( 2001 ) .
z. b. etienne , y. t. liu , s. l. shapiro , t .
w. baumgarte , phys .
d * 79 * , 044024 ( 2009 ) .
e. rantsiou , s. kobayashi , p. laguna , f. a. rasio , astrophys .
j. * 680 * , 1326 , ( 2008 ) .
m. vallisneri , phys .
lett . * 84 * , 3519 ( 2000 ) .
m. shibata , prog .
. phys . * 96 * , 917 ( 1996 ) .
j. a. faber , class .
* 26 * , 114004 , ( 2009 ) .
j. a. faber , t. w. baumgarte , s. l. shapiro , k taniguchi , f. a. rasio , phys .
d * 73 * , 024012 ( 2006 ) .
j. a. faber , t. w. baumgarte , s. l. shapiro , k taniguchi , astrophys .
j. * 641 * , l93 , ( 2006 ) .
m. shibata , k. uryu , class .
. grav .
* 24 * , s125 , ( 2007 ) .
k. taniguchi , t. w. baumgarte , j. a. faber , s. l. shapiro , phys .
d * 75 * , 084005 , ( 2007 ) .
s. chandrasekhar _ ellipsoidal figures of equilibrium _ ( the silliman foundation lectures , new haven : yale university press ) , ( 1969 ) . j .- a . marck proc .
a , * 385 * , 431 , ( 1983 ) .
w. h. press , s. a. teukolsky , w. t. vetterling , and b. p. flannery , _ numerical recipes in fortran .
the art of scientific computing _
( cambridge university press , 1992 ) .
lattimer , m. prakash , phys . rep . * 442 * , 109 , ( 2007 ) .
j. m. bardeen , w. h. press , s. a. teukolsky , astrophys .
j. * 178 * , 347 , ( 1972 ) .
a. buonanno , y. chen , and t. damour , phys .
d * 74 * , 104005 ( 2006 ) .
g. schfer , ann . phys . * 161 * , 81 ( 1985 ) .
r. b. wiringa , v. g. j. stoks , r. schiavilla , phys .
c * 51 * , 38 , ( 1995 ) .
b. s. pudliner , v. r. pandharipande , j. carlson , s. c. pieper , r. b. wiringa , phys . rev .
c * 56 * , 1720 , ( 1997 ) .
a. akmal , v. r. pandharipande , phys .
c * 56 * , 2261 , ( 1997 ) .
a. akmal , v. r. pandharipande , d.g .
ravenhall phys .
c * 58 * , 1804 , ( 1998 ) .
r. schiavilla , v. r. pandharipande , r.b .
wiringa , nucl .
a449 * , 219 ( 1986 ) .
m. baldo , g. f. burgio , h. j. schulze , phys .
c * 61 * , 055801 , ( 2000 ) .
s. balberg , a. gal , nucl .
phys . a * 625 * , 435 , ( 1997 ) .
j. m. lattimer , f. d. swesty , nucl .
phys . a * 535 * , 331 , ( 1991 ) .
n. k. glendenning , astrophys .
j. * 293 * , 470 , ( 1985 ) .
i. bombaci , in perspectives on theoretical nuclear physics , editors : i. bombaci et al .
, ets pisa , 222 ( 1996 ) ; + m. prakash , i. bombaci , m. prakash , p. j. ellis , j. m. lattimer , r. knorren , phys
. rep . * 280 * , 1 ( 1997 ) .
f. douchin , p. haensel , phys .
b * 485 * , 107 ( 2000 ) .
f. douchin , p. haensel , astron . & astroph .
* 380 * , 151 ( 2001 ) .
p. haensel , b. pichon , astron .
& astroph .
* 283 * , 313 ( 1994 ) .
g. baym , c. pethick , p. sutherland , astrophys .
j. * 170 * , 299 , ( 1971 ) .
k. g. arun , b. r. iyer , b .
s. sathyaprakash , p. a. sundararajan , phys .
d * 71 * , 084008 , ( 2005 ) .
b. s. sathyaprakash , b. f. schutz , living reviews in relativity * 12 * , 2 , ( 2009 ) . | we study the tidal disruption of neutron stars in black hole - neutron star coalescing binaries .
we calculate the critical orbital separation at which the star is disrupted by the black hole tidal field for several equations of state describing the matter inside the neutron star , and for a large set of the binary parameters .
when the disruption occurs before the star reaches the innermost stable circular orbit , the gravitational wave signal emitted by the system is expected to exhibit a cutoff frequency @xmath0 , which is a distinctive feature of the waveform .
we evaluate @xmath0 and show that , if this frequency will be found in a detected gravitational wave , it will allow one to determine the neutron star radius with an error of a few percent , providing valuable information on the behaviour of matter in the stellar core . |
the large hadron collider ( lhc ) will measure the @xmath12 jet production cross section in hadron collisions for various ranges of transverse momenta ( @xmath5 ) with unprecedented statistics and kinematics . with the improvements of @xmath12 tagging pioneered at the tevatron @xcite , the analysis of @xmath12 production with associated muon tags in the lhc environment @xcite is very promising .
studies by the cms collaboration have determined that the final states of @xmath12 production can be triggered on and measured out to @xmath4 in the @xmath5 of inclusive @xmath6 meson production by utilizing @xmath12 jet tagging in events containing at least one muon @xcite . the standard model ( sm )
qcd production mechanism for @xmath3 falls rapidly when @xmath13 , thus , studies with restricted @xmath1 ( for such values of @xmath5 and a pseudorapidity ( @xmath14 ) of @xmath2 the variables @xmath5 and @xmath14 are independent from one another ) offer the opportunity to explore the effect that beyond the sm ( bsm ) physics can have on the @xmath5 tail of the @xmath15 distribution . in utilizing this distribution to search for bsm physics one
has to contend with the fact that nlo qcd production leads to a larger background than lo qcd at high @xmath5 due to a @xmath16-channel singularity in @xmath17 . in this paper
, we argue that the nlo qcd production background can be significantly reduced through simple kinematic cuts .
when the cuts are imposed , the dominant remaining qcd production processes are @xmath10 and bsm physics can more easily produce statistically significant excesses of high @xmath5 events .
we restrict our attention to examples of bsm physics with massive new octets under @xmath7 and find that measurements of @xmath15 are quite promising to constrain some @xmath7 octets , affording a @xmath18 discovery reach out to @xmath19 using these cuts .
the appearance of new bosons of the form discussed in this paper can be motivated in many scenarios of bsm physics .
we concentrate on two models where they appear : a model when a larger gauge group breaks down to qcd as a diagonal subgroup @xcite and the manohar - wise model @xcite which considers minimal flavor violation ( mfv ) @xcite as a global symmetry of new physics that allows specific scalar representations @xcite .
we first discuss isolating the signal region by suppressing the nlo qcd production mechanisms through simple kinematic cuts and determine analytically the resulting lo qcd contribution to @xmath15 .
we then examine the effect on the calculated distribution due to spin one @xmath7 octets and determine the @xmath18 discovery reach .
quarks in both cases can be related through common perturbative fragmentation functions to @xmath6 mesons or other hadronic final states ( or jet functions ) and the relative event rates and discovery reach conclusions will not be significantly effected by this further complication . ] the tree level effects of some new @xmath7 bosons can be small .
this is the case if the strength of the boson s coupling to quarks is proportional to the quark mass . in some models , the largest effect of new
physics comes through loop corrections interfering with the sm production mechanisms @xmath10 or one - loop production that is enhanced with an @xmath9-channel resonance .
we explore this possibility in detail when the sm is supplemented with an @xmath11 scalar motivated by mfv @xcite .
however , for this model , the potential for discovery in this channel is not strong .
the dominant contribution to the @xmath15 distribution is qcd production .
we first discuss our estimates of the qcd production mechanisms and the isolation of this signal before comparing the bsm signal to this background .
when kinematic cuts that we advocate are imposed , the dominant sm production mechanisms are @xmath10 and the bsm signature must be detectable above this background .
the hadronic momenta are defined as @xmath20 , where @xmath21 is the hadronic center of mass energy .
the momentum of the @xmath12 quark is defined in terms of @xmath22 by p = ( |p_t| , _
t , |p_t| ) .
[ loqcd ] for qcd , the production processes @xmath10 begin at @xmath23 ( see figure 1 ) .
alternate heavy flavor production mechanisms through gluon splitting @xmath24 and direct heavy flavor excitation @xmath25 begin at @xmath26 .
the contributions from gluon splitting and flavor excitation are dominant when all phase space is integrated over .
this has been appreciated theoretically for some time , see @xcite .
studies of the total cross section at lhc ( operating at @xmath27 ) using pythia @xcite find that @xmath0 is approximately @xmath28 , with gluon splitting and heavy flavor excitation processes contributing @xmath29 each while the processes @xmath10 contribute @xmath30 .
this is mostly due to the @xmath16-channel rutherford scattering enhancement at large @xmath5 for @xmath31 and the effect of pdfs ; @xmath32 is approximately a factor of 200 larger than @xmath33 in the high @xmath5 region .
the different processes can be distinguished through their final state kinematics .
the processes @xmath10 yield @xmath3 pairs that tend to be back to back and highly symmetric in @xmath5 . in flavor excitation ,
only one of the quarks undergoes a hard scattering event and the final state @xmath3 pairs are highly asymmetric in @xmath5 . in gluon
splitting , the @xmath3 pairs are dominantly produced with an opening angle due to splitting and neither of the quarks undergos a hard scattering interaction .
this typically leads to @xmath3 pairs with a small opening angle and small @xmath5 , or both with large @xmath5 but not back to back . using cuts ,
the different processes can be separated out and a dominant region of @xmath10 production can be isolated .
one strategy is to use muonic tags of the @xmath3 decays from the semileptonic @xmath34 decays and @xmath35 decays to allow triggering and to improve the identification and discrimination of @xmath12 jets from backgrounds .
a proof of principle is given by an initial cms study that utilizes the opening angle between the @xmath12 quarks to distinguish production mechanisms @xcite .
this initial study demonstrates the promise of discriminating the various production mechanisms in a hadron collider environment .
note that @xcite insisted on a single @xmath35 decay as it was interested primarily in identifying states with small opening angles .
we are interested in @xmath12 jets that are approximately back to back with semi - leptonic muon decays .
we advocate a discrimination of the production mechanisms through identified @xmath12 jets by enforcing cuts on the @xmath5 asymmetry and the pseudorapidities of the reconstructed @xmath3 pair of the following form , * to isolate the signal region with a small @xmath5 asymmetry enforce @xmath36 where @xmath37 is the total error on the reconstruction of the quark s partonic @xmath5 .
the particular value of @xmath38 is not required for our study as the lo production processes we consider trivially satisfy this condition .
an optimal value of @xmath38 can be obtained in a nlo study .
the error for the @xmath5 asymmetry depends on the decay products and the reconstruction algorithm used ; for the muonic decay tags that we advocate , a cms study @xcite finds a reconstruction error of @xmath39 on @xmath5 of the @xmath6 hadron and fragmentation function errors on the total cross section are on the order of @xmath40 .
this cut efficiently reduces the nlo flavor excitation production . *
the background due to gluon splitting can be reduced through insisting that the pseudorapidities of the @xmath12 jets lie on different sides of the beam line in the polar angle .
this can be done through imposing a cut with @xmath41 on a nlo calculation .
the produced @xmath42 pairs will not lie exactly back to back at lo due to the residual boost resulting from the asymmetry of the partonic momenta via pdf effects .
these conditions are trivially satisfied by the two - body bsm or qcd production processes for the @xmath5 cuts that we consider while efficiently suppressing the large nlo background from gluon splitting and flavor excitation .
further evidence in support of the efficiency of these analytic cuts in isolating the signal region at lhc is supplied by past monte carlo studies of the kinematic properties of the various heavy flavor creation processes at the tevatron @xcite .
we compare the effect of bsm bosons on @xmath3 production in the regions with the highest @xmath5 that should be accessible in the cms detector .
we restrict the @xmath5 of the final state @xmath3 quarks to generally lie in the range @xmath43 .
the pseudorapidity range is the full geometric acceptance of the cms detector for @xmath2 @xcite .
it turns out that for such values of @xmath5 and @xmath14 those two quantities are independent from each other and we can impose separate cuts on them . and @xmath5 it is sufficient to take @xmath44 . for @xmath2
this yields @xmath45 . ]
the differential cross section of the qcd and bsm processes in terms of the spin and color averaged matrix element @xmath46 ( in terms of partonic mandelstam variables ) is given by [ eq ] = _ 0 ^ 1 d x_2 _ 0 ^ 1 d x_1 f(x_1 , ) f(x_2 , ) .
the @xmath47 are the pdfs evaluated at the renormalization scale @xmath48 . as usual , @xmath49 relate the partonic and hadronic momenta so that @xmath50 . integrating over pseudorapidity , the relevant differential cross section for our @xmath5 cuts is given by = _
^1 d x_2 _ ^1 d x_1 f(x_1 , ) f(x_2 , ) .
the lo qcd @xmath51 amplitudes are given in appendix a , section 1 .
the results for various integrated luminosities , @xmath5 ranges , and cm energies for the lhc are given in table i. @xmath52 & lo [ @xmath53 , 1.4 ] & lo [ 1.0 , 1.4 ] + & 200 @xmath54 & 68 & 4.4 + [ 0pt]7 & 10 @xmath55 & @xmath56 & 220 + 10 & 10 @xmath55 & @xmath57 & @xmath58 + 14 & 10 @xmath55 & @xmath59 & @xmath60 + the contributions of the two - body qcd production mechanism for @xmath0 up to nlo are known @xcite . in comparing our bsm results with the sm qcd background
we restrict ourselves to lo qcd for this production process .
we also restrict our attention to lo qcd due to collinear singularities in the nlo qcd production processes that are regulated by @xmath61 .
we present results of one - loop interference calculations between the sm and bsm physics where the @xmath12 quark mass is neglected . to compare these results consistently with nlo qcd calculations
we would need to include final state masses in both calculations as well as contributions from processes with arbitrarily soft extra gluon emission .
the dominant nlo qcd production mechanisms that are larger than the lo qcd production mechanisms ( flavour excitation and gluon splitting ) that we are concerned with are suppressed by the cuts we advocate .
however , nlo qcd perturbative corrections of the form of virtual corrections and initial and final state radiation corrections to the lo qcd production rate , that do not have these distinct kinematic signatures , will not be highly suppressed by these cuts . including such corrections
is beyond the scope and purpose of this work .
we treat the spectrum of the bsm physics as an eft with a single @xmath62 singlet or doublet of bosons added to the sm in each case , assuming the cut - off scale is sufficiently high compared to this mass so that higher dimensional operator effects can be neglected .
we begin with the case with tree level exchanges of a massive new spin one @xmath7 boson with a purely vector coupling , sometimes referred to as a coloron in the literature .
consider supplementing the sm with a massive spin one @xmath7 octet within the spectrum of bsm physics .
many models of dynamical electroweak symmetry breaking with new strong interactions have such a particle , for details see @xcite . a sufficient condition for this particle to appear in a bsm spectrum
is to have a symmetry breaking pattern where a higgs mechanism is employed for qcd to emerge as a diagonal subgroup @xmath63 . when the resultant spectrum is reduced to its mass eigenstate basis one obtains @xmath64 with the addition of a massive @xmath7 octet .
gauge invariance forbids the single coupling of such a particle to gluon pairs and pair production leading to four @xmath12 and four jet signatures of these states has been recently examined in @xcite .
consider the case where the coupling of the massive @xmath7 octet has purely vector - like flavor universal coupling and @xmath65 , where @xmath66 is the ratio of the gauge couplings of the underlying @xmath67 gauge groups .
then single production of these states from quarks in @xmath68 collisions will give rise to the dominant source of @xmath3 pairs with the kinematics we are interested in .
the spin and color averaged matrix element for the excess contribution compared to qcd in the @xmath9-channel in terms of partonic mandelstam variables is ||^2= ( ) ^4 c_f^2 ( + ) , where the width of the coloron is approximated by its quark decays as @xmath69 . for colorons of mass @xmath70
the excess of events in the @xmath5 tail is far larger than the uncertainties of the lo qcd estimate of the background .
we define a @xmath18 significance by treating the @xmath71 qcd rate as a poisson distribution . to estimate the effect of large @xmath72 corrections we multiply the @xmath71 qcd rate used in the poisson distribution by @xmath73 $ ] and we take @xmath74 .
qcd calculation utilizing the @xmath75 formalism of @xcite which resums the large logs of @xmath76 but this is beyond the scope of this work .
such a resummation while running down from the @xmath5 scale to @xmath61 justifies our choice of the renormalization scale at @xmath77 . ]
we insist on five signal events after a suppression factor of @xmath78 is applied for the branching ratios of both @xmath12 quarks to final states containing a muon and a total signal efficiency of @xmath79 @xcite for each muon , @xmath12 jet trigger .
a @xmath18 discovery of colorons is possible in this channel with a reach out to @xmath80 , however , a significant event rate will require @xmath81 of integrated luminosity because of the suppression of the signal due to selection efficiencies and branching ratios .
we show the total integrated luminosity required for various coloron masses and operating energies that pass these tests in figure 2 .
other sources of events from the @xmath82 background in the signal region are estimated to be a small @xmath83 level background @xcite .
fakes from @xmath84 quarks are reduced with the associated @xmath12 jet trigger and light quark fakes are removed by a @xmath85 cut on the muons .
the experimental cross section uncertainty is approximately @xmath86 @xcite for the @xmath5 regions we consider and is dominated by a systematic jet energy scale uncertainty of @xmath87 .
[ loqcd ] tail of @xmath3 production for @xmath5 in the range @xmath88 for cm energies @xmath89 , @xmath90 , and in the range @xmath91 to @xmath92 for @xmath93 .
we show the range of coloron masses that can be discovered with @xmath18 significance and at least 5 signal events after branching ratio and event selection suppression of @xmath94 is applied to the signal rate . on the left we take @xmath95 , on the right @xmath96 .
our discovery reach is dominated by the requirement of @xmath97 signal events after the suppression of @xmath94 is applied to the signal rate , not pdf or theory uncertainties .
for example , the signal to background ratio for the entire discovery reach region for @xmath98 and @xmath99 is @xmath100.,title="fig:",height=204 ] some of the coloron masses which the tail of the @xmath5 distribution is sensitive to are not ruled out by tevatron studies . the most stringent and most model - independent bounds on the mass scale of flavor universal colorons come from resonance searches in the di - jet invariant mass differential distribution @xcite which obtains a @xmath101 confidence limit exclusion bound on the coloron masses in the range of @xmath102 .
we emphasize the complementary nature of searches in this channel to the traditional search strategies for colorons that have focused on @xmath103 pair production .
new @xmath7 bosons with tree level couplings to quarks , where @xmath104 , can lead to a large excess of events in the tail of the @xmath8 distribution . for small widths a resonance peak can also be reconstructed . however ,
when the tree level coupling of a new boson to quarks is proportional to the quark mass , which is frequently the case for spin zero bosons , the largest contribution to high @xmath5 events can come from one - loop production interference effects with the sm lo production processes .
this is the case for a @xmath11 scalar representation that couples to quarks proportionally to the yukawa coupling . in this section
we explore the possible effects on the tail of the @xmath8 distribution due to such a scalar in one - loop corrections .
one - loop effects should not be assumed too small to be detected for heavy flavor production @xmath105 due to our kinematic isolation of a signal region with a suppressed qcd background in the tail of the steeply falling @xmath8 distribution .
a direct example of the need of considering naively higher order corrections that experience kinematic enhancements is given by the dominance of nlo qcd corrections over lo qcd in heavy flavor production when all phase space is integrated over .
in this particular model , two classes of corrections are of potential interest : a class of one - loop corrections to @xmath106 production that are proportional to an enhanced top yukawa coupling and a separate class experiencing a resonant enhancement in the @xmath9-channel .
we investigate these loop effects in detail in the following subsection . in the manohar - wise model
@xcite one adds a single @xmath11 scalar representation to the sm .
this representation is the only flavor singlet @xcite scalar representation allowed by mfv other than the higgs .
we denote the doublet as ^a = ( c s^+^a + s^0^a ) , where @xmath107 is the color index . the yukawa sector is determined up to overall complex normalization constants , @xmath108 and @xmath109 , as required by mfv to be l = -_u g_i j^u |u_r i t^a q_j s^a-_d g_i j^d |d_r i t^a q_j s^a + h.c .
, where @xmath110 and @xmath111 are the standard model yukawa matrices and @xmath112 are flavor indices .
see @xcite for a discussion on the current status of the model s phenomenology .
the largest effects of this model on the tail of the @xmath8 distribution do not come from tree level exchanges .
this is due in part to the suppression by light quark yukawa couplings .
the @xmath113 quark yukawa couplings are far larger but their anti - quark pdfs are highly suppressed in the proton .
further , gauge invariance forbids a coupling of a single octet scalar to two gluons .
the largest effects on @xmath8 come from two potential sources : one - loop production mechanisms that afford an @xmath9-channel resonance with an @xmath114 exchange , and one - loop interference terms with the qcd production mechanisms which are proportional to @xmath115 due to @xmath116 loops .
we discuss the effect of each of these on the tail of the @xmath8 distribution in turn .
the largest @xmath9-channel resonant production is shown in figure 3 .
scalars.,title="fig:",height=207 ] from the loop parts of those diagrams , the biggest effect on single production of neutral scalars comes from the top quark loop @xcite .
this is also true for the @xmath9-channel exchange and we neglect the effects of the other diagrams that are also proportional to the unknown couplings in the scalar potential of the model ( @xmath117 ) , which are expected to further suppress the contribution of the scalar loop diagrams .
the amplitude for the top loop is given by i _ s = _ 0 ^ 1 dx _ 0 ^ 1-x dy where we have taken @xmath14 s real and neglected terms proportional to @xmath118 as discussed in the appendix .
we approximate the width of the real scalar by @xmath119 as it is dictated by top decays @xcite .
we have determined @xmath120 and used equation ( 3 ) to calculate the resulting contribution of the excess events in the @xmath5 tail . unlike the sm higgs , as the @xmath121 is an 8 under @xmath7 , the amplitude in equation ( 7 ) interferes with the sm heavy flavor production amplitudes given in appendix a , section 1 .
this process leads to the change in the number of events given by @xmath122 in table ii and the interference introduces another factor of the quark mass .
the resulting event rate is too small to be observed because the resonant enhancement is suppressed by @xmath123 and loop factors .
quark final states to have an observable event rate in the @xmath5 tail of the @xmath124 distribution .
] for large masses the parameters @xmath125 can be greater than one , however , the constraints from this model s loop corrections to @xmath126 and @xmath127 ( see , @xcite ) still constrain @xmath128 , so no statistically significant excess is observable in the @xmath5 tail from these processes with final state @xmath12 quarks .
the largest non - resonant effect comes from the interference of one - loop corrections due to @xmath129 with the sm lo production processes .
these corrections are proportional to @xmath115 .
the resulting amplitudes and diagrams are given in the appendix .
again , after a rather lengthy calculation we have determined the excess event rate due to these effects using equation ( 3 ) .
we summarize in table ii the numbers of expected excess events for the various loop processes .
.the change in the number of @xmath3 events due to interference with the octet amplitudes where we restrict @xmath5 between 1.0 and 1.4 @xmath130 , and use an operating energy of @xmath27 for lhc .
the results in the table do not have a selection cut or branching ratios to muons applied .
the results are given using mstw pdfs for @xmath131 of integrated luminosity .
we use @xmath132 , @xmath133 and take @xmath134 .
the process for @xmath135 is given in appendix a , section 2 .
the process for @xmath136 is given in appendix a , section 3 .
we have checked that the processes for @xmath135 and @xmath137 analytically have the correct decoupling behavior .
the remaining two processes come from the direct square and the interference of equation ( 7 ) with the lo qcd results in appendix a , section 1 .
these are negligible despite the resonance enhancement .
we present the results with the unknown @xmath108 and @xmath109 factored out .
we also show the constraints on these parameters from allowing a @xmath138 deviation of @xmath127 @xcite and a @xmath139 deviation in @xmath126 @xcite .
light octets are not ruled out by direct production and are consistent with ewpd down to @xmath140 @xcite .
all of these mechanisms do not lead to a statistically significant excess of events in the @xmath5 tail of @xmath42 production . [
cols="^,^,^,^,^,^",options="header " , ] for the @xmath5 range of integration @xmath141 the results are not large enough to be detected , despite the highly suppressed background and the enhancement of these loop effects .
the lo qcd background to compare these results to is given in table i. the experimental uncertainty of the cross section in the high @xmath5 region has a systematic uncertainty on the order of @xmath142 .
no statistically significant excess is expected in this model in the @xmath5 tail .
our results additionally imply that these effects on the @xmath3 cross section determined at the tevatron are also negligible , affording no significant constraint on the model .
heavy flavor production at hadron colliders has a long history of surprises and challenges to both theorists and experimentalists . due to the long standing difficulties of accommodating the measured @xmath143 at the tevatron within our understanding of qcd , which have only recently been overcome , exploring high @xmath5 production of @xmath42 pairs at the lhc as a constraint on bsm physics demands great caution . despite this
, one can impose simple kinematic cuts to improve the discovery reach of heavy flavor production by suppressing the nlo qcd flavor excitation and gluon splitting processes .
this allows one to more efficiently exploit @xmath12 tagging with associated muons which affords a significant @xmath5 reach for final state @xmath6 mesons at lhc .
we have determined the reach for a @xmath144 discovery of flavor universal colorons using this technique at lo in qcd .
we have also demonstrated that the effects on @xmath3 production in the manohar - wise model are not large enough to afford a statistically significant deviation from the expected qcd production background . for the @xmath5
reach in @xmath8 to be fully exploited in the lhc era , further work on precision studies of heavy flavor production in the standard model is of increased importance in light of this discovery potential
. improvements of the sm calculations of heavy flavor production and the associated systematic uncertainties would have to be carefully examined at lhc before any evidence of new physics could be claimed .
however , our results are promising for further work to develop the heavy flavor production program at lhc with this aim for discovery of new massive @xmath7 bosons with tree level couplings to quarks not proportional to the quark mass
. we thank mark wise for collaboration during the early stages of the work presented here .
we further thank mark wise , maxim pospelov , saba zuberi and brian batell for comments on the manuscript .
research at the perimeter institute is supported in part by the government of canada through nserc and by the province of ontario through medt .
the lo and nlo qcd results for @xmath0 are well known in the literature @xcite .
we summarize the simple lo results for completeness and to explain a trick @xcite that we employ to simplify our subsequent one - loop calculations . as discussed in section ii , cuts on the @xmath5 asymmetry and the opening angle of the reconstructed @xmath3 pair
can isolate the kinematic region where the dominant qcd ( or bsm ) production mechanisms are the @xmath10 processes .
the lo results for these processes in qcd are given by @xmath145 \ , \left[\bar{u}(p ) \
, \gamma_{\mu } \ , t^a \ , v(\bar{p } ) \right ] , \nonumber\\ i \mathcal{m}^{gg}_{sm}&=&-i g_s^2\ , \bar{u}(p)\bigg[[t^a , t^b]\frac{1}{s } \left(g^{\mu\nu}(\slashed{p}_1-\slashed{p}_2)+2\gamma^{\nu}p_2^{\mu}-2\gamma^{\mu}p_1^{\nu}\right)\nonumber\\ & & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ + \,t^bt^a\frac{1}{t } \gamma^{\nu}(\slashed{p}_1-\slashed{\bar{p}})\gamma^{\mu}+ t^at^b\frac{1}{u}\gamma^{\mu}(\slashed{p}-\slashed{p}_1)\gamma^{\nu}\bigg ] v(\bar{p})\ .\end{aligned}\ ] ] note that in @xmath146 we have modified the triple gluon vertex of qcd to eliminate the dependence on @xmath147 .
this does not change the final answer once the gluon is summed over physical polarizations as the difference is proportional to @xmath148 and @xmath149 @xcite .
this trick does , however , allow us to sum over the physical and non - physical polarizations of the gluon as the spurious contributions to the amplitude , which are proportional to this momentum dependence , are removed .
when calculating the interference of the lo qcd processes with such a modification , this trick allows one to set to zero all factors of @xmath147 in our results for the one - loop interference effects of the @xmath11 scalar with qcd , and sum over all polarizations of the gluon , simplifying significantly intermediate steps of the calculation .
our one - loop result for the modification of @xmath151 ( see figure 4 ) is given as the following gauge invariant amplitude ( where we have set all external masses to zero ) @xmath152 \ , \left[\bar{v}_i(p_2 ) \ , \gamma_\mu \ , t^a u_i ( p_1 ) \right]\nonumber\\ & & \times \bigg[c_a \bigg [ a_1(m_s^2,m_t^2 ) -a_1(m_t^2,m_s^2 ) + a_2(m_s^2,m_t^2)+ \frac{1}{2 } \bigg]\nonumber\\ & & + \,\,c_f \bigg[\ , 2 \ , a_1(m_t^2,m_s^2 ) - \,2 \ , a_2(m_s^2,m_t^2 ) + \frac{3m_t^2-m_s^2}{2(m_s^2-m_t^2)}\,\bigg]\nonumber\\ & & + \,\,c_f\bigg[\,\frac{m_s^2(m_s^2 - 2m_t^2)}{(m_s^2-m_t^2)^2}\ln\left(\frac{m_s^2}{\mu^2}\right ) + \frac{m_t^4}{(m_s^2-m_t^2)^2}\ln\left(\frac{m_t^2}{\mu^2}\right ) \bigg]\bigg ] \ , \end{aligned}\ ] ] where a_1(m_s^2,m_t^2 ) & = & _ 0 ^ 1 dx ( ) + & & -_0 ^ 1 dx ( ) , + a_2(m_s^2,m_t^2 ) & = & - _ 0 ^ 1 dx + ( ) + & & + _ 0 ^ 1 dx ( ) , and for the @xmath7 case @xmath153 .
the formulas for the @xmath158-channel diagrams are obtained from the @xmath16-channel results by substituting @xmath159 .
we define the following feynman parameter volumes as a short hand d v(x , y ) & = & _ 0 ^ 1 dx _ 0 ^ 1-x dy , + d v(x , y , z ) & = & _ 0 ^ 1 dx _ 0 ^ 1-x dy _ 0 ^ 1-x - y dz .
our @xmath16-channel results are as follows ( again , after setting all external masses to zero ) .
the expressions below should be contracted with @xmath160 .
@xmath161x t\bigg]\ , \nonumber\\ \mathcal{m}^t_{e}&= & \frac { c_a \ , t^b \ , t^a}{t } \ , \bigg[\bar{u}_l(p ) \gamma_{\nu}(\slashed{p}_1-\slashed{\bar{p}})\gamma_{\mu}v_l(\bar{p } ) \ , \int d v(x , y ) \ , \ln\left(\frac { f_2(x , y)}{\mu^2}\right)\nonumber\\ & & + \ , \int d v(x , y ) \ , \frac{1}{f_2(x , y ) } \bar{u}_l(p)\gamma_{\mu}v_l(\bar{p})[(p_1-\bar{p})_{\nu}(2x-1)+p_{\nu}(2y-1)]x t\bigg]\ , \nonumber\\ \mathcal{m}^t_{f}&= & \frac{\left ( 2 \ , c_f - c_a\right)t^bt^a}{t } \bigg[\bar{u}_l(p ) \gamma_{\nu}(\slashed{p}_1-\slashed{\bar{p}})\gamma_{\mu}v_l(\bar{p})\ , \int d v(x , y ) \ , \left [ \ln \left(\frac{f_{3}(x , y)}{\mu^2}\right)+1\right ] \nonumber\\ & & - \int d v(x , y ) \ , \frac{1 } { f_{3}(x , y)}\bar{u}_l(p ) \gamma_{\nu}(\slashed{p}_1-\slashed{\bar{p } } ) [ \slashed{p}_1(1-x)-\slashed{\bar{p}}y+m_t]\gamma_{\mu}[-\slashed{p}_1x - \slashed{\bar{p}}y + m_t]v_l(\bar{p } ) \bigg]\ , \nonumber\\ \mathcal{m}^t_{g}&= & \frac{\left(2 \ , c_f - c_a\right)t^bt^a}{t } \bigg
[ \bar{u}_l(p ) \gamma_{\nu}(\slashed{p}_1-\slashed{\bar{p}})\gamma_{\mu}v_l(\bar{p})\int d v(x , y ) \ , \left[\ln \left(\frac{f_{4}(x , y)}{\mu^2}\right)+1\right]\nonumber\\ & & - \int d v(x , y ) \ , \frac{1 } { f_{4}(x , y)}\bar{u}_l(p ) [ \slashed{p}_2x+\slashed{p}y+m_t]\gamma_{\nu}[\slashed{p}_2(x-1)+\slashed{p}y+m_t ] ( \slashed{p}_1-\slashed{\bar{p}})\gamma_{\mu}v_l(\bar{p } ) \bigg]\ , \nonumber\\ \mathcal{m}^t_{h}&=&0\ , \end{aligned}\ ] ] where @xmath162 the @xmath9-channel and @xmath16-channel box diagram results are @xmath163 - \bar{u}_l(p)\gamma_{\mu}v_l(\bar{p})\bigg[\frac{p_{\nu}(1-x - y)+\bar{p}_{\nu}z}{2f_{5}(x , y , z)}\bigg ] \nn \\ & & + \,\bar{u}_l(p)\slashed{p}_1v_l(\bar{p})\bigg [ \frac{x g_{\mu\nu}}{2f_{5}(x , y , z)}+\frac{x\left[\bar{p}_{\mu}(1-x - z)+p_{\mu } y\right ] \left[p_{\nu}(1-x - y)+\bar{p}_{\nu}z\right]}{[f_{5}(x , y , z)]^2}\bigg ] \ , \bigg ] , \nn \\
! \left[(2 c_f - c_a)t^bt^a + \frac{c_a}{2}\delta^{ba}\right]\!\int \!d v(x , y , z)\bigg [ \frac{1}{f_{6}(x , y , z)}\bar{u}_l(p ) \gamma_{\nu}\gamma_{\mu}[\slashed{p}_1(y+z-1)-\slashed{p}_2 x ] v_l(\bar{p } ) \bigg . \nn
\\ & & \bigg.+\frac{1}{f_{6}(x , y , z)}\bar{u}_l(p)\big[\gamma_{\mu}[\slashed{p}_1 z - \slashed{p}_2 ( x+y)+\slashed{p}y]\gamma_{\nu}+ [ \slashed{p}_1 z + \slashed{p}_2(1-x - y)]\gamma_{\nu } \gamma_{\mu}\big]v_l(\bar{p } ) \bigg . \nn \\ & & \bigg.+ \frac{1}{[f_{6}(x , y , z)]^2 } \bar{u}_l(p ) m_t^2 \big [ [ \slashed{p}_2(1-x - y)+\slashed{p}_1 z]\gamma_{\nu}\gamma_{\mu } + \gamma_{\nu}[-\slashed{p}_2 ( x+y ) + \slashed{p}_1 z + \slashed{p } y]\gamma_{\mu}\big]v_l(\bar{p } ) \bigg .
\nn \\ & & \bigg.+ \frac{1}{[f_{6}(x , y , z)]^2 } \bar{u}_l(p ) m_t^2\gamma_{\nu}\gamma_{\mu}[-\slashed{p}_1 ( 1-y - z)-\slashed{p}_2 x]v_l(\bar{p } ) \bigg .
\nn \\ & & \bigg.\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!+ \frac{1}{[f_{6}(x , y , z)]^2 } \bar{u}_l(p ) [ \slashed{p}_2(1-x - y)+\slashed{p}_1 z ] \gamma_{\nu}[-\slashed{p}_2(x+y)+\slashed{p}_1 z + \slashed{p}y]\gamma_{\mu}[-\slashed{p}_2 x - \slashed{p}_1(1-y - z ) ] v_l(\bar{p } ) \bigg ] , \nonumber\\ \mathcal{m}^t_{k}&= & - 8 \ , c_a \ , \delta^{ab } \int d v(x , y , z ) \bigg [ \frac{1}{f_{7}(x , y , z ) } \ , \bar{u}_l(p ) \ , \gamma_{\mu}[p_2^{\nu}(2y+2z-1)-2\bar{p}^{\nu}y+2p^{\nu}x ] v_l(\bar{p } ) \bigg .
\nn \\ & & + \bigg .
\,\frac{1}{f_{7}(x , y , z ) } \ , \bar{u}_l(p)\big[\gamma_{\nu}\gamma_{\mu}[-\slashed{p}_2(y+z-1)-\slashed{p}x+m_t ] + [ -\slashed{p}_2(y+z)+\slashed{\bar{p}}y+m_t]\gamma_{\mu}\gamma_{\nu}\big]v_l(\bar{p } ) \bigg .
\nn \\ & & + \,\frac{1}{[f_{7}(x , y , z)]^2}\bar{u}_l(p)[-\slashed{p}_2(y+z)+\slashed{\bar{p}}y+m_t]\gamma_{\mu } [ -\slashed{p}_2(y+z-1)-\slashed{p}x+m_t][2\bar{p}^{\nu}y-2p^{\nu}x]v_l(\bar{p } ) \bigg ] , \nonumber\\ \mathcal{m}^s_{l , m}\!&=&\!\!-\frac{2 c_f}{s } \ , \left[t^a , t^b\right ] \bar{u}_l(p)\left[g^{\mu\nu } ( \slashed{p}_1-\slashed{p}_2)+2\gamma^{\nu}p_2^{\mu}-2\gamma^{\mu}p_1^{\nu}\right]v_l(\bar{p } )
\int_0 ^ 1 dx \ , x \
, \ln \left(\frac{m_s^2x+m_t^2(1-x)}{\mu^2}\right),\nonumber\\ \mathcal{m}^s_{n}&=&\frac{c_a}{s } \left[t^a , t^b\right ] \bar{u}_l(p)\left[g^{\mu\nu } ( \slashed{p}_1-\slashed{p}_2)+2\gamma^{\nu}p_2^{\mu}-2\gamma^{\mu}p_1^{\nu}\right]v_l(\bar{p } ) \int d v(x , y ) \
, \ln \left(\frac{f_8(x , y)}{\mu^2}\right ) , \nonumber\\ \mathcal{m}^s_{o}&= & \frac{\left ( 2 c_f - c_a\right)}{s } \ , \left[t^a , t^b\right ] \ , \bar{u}_l(p)\left[g^{\mu\nu } ( \slashed{p}_1-\slashed{p}_2)+2\gamma^{\nu}p_2^{\mu}-2\gamma^{\mu}p_1^{\nu}\right]v_l(\bar{p } ) \nn \\ & & \times \int d v(x , y ) \ , \bigg[\ln \left(\frac{f_9(x , y)}{\mu^2}\right)+1- \frac{m_t^2 + s x y } { f_9(x , y)}\bigg]\ , \!\!\!\!\!\!\!\!\!\!\!\!\!\end{aligned}\ ] ] where @xmath164 | we discuss the potential of measurements of @xmath0 to constrain new bosonic degrees of freedom at the lhc when @xmath1 for a pseudorapidity cut @xmath2 . by suppressing the nlo qcd production of @xmath3 pairs through simple kinematic constraints
we show how to more efficiently exploit cms s reach out to @xmath4 in @xmath5 for @xmath6 mesons in searches for new physics . using this technique
we investigate the potential for discovery of new massive spin one and spin zero @xmath7 octet bosons by analyzing their effect on the @xmath8 distribution . in some cases , the tree level contributions are small and the largest effect of new physics on the @xmath5 tail of the distribution is due to one - loop resonant @xmath9-channel production or interference effects with the lo standard model production mechanisms @xmath10 .
we explore this possibility in some detail when the standard model is extended with an @xmath11 scalar motivated by minimal flavor violation . |
the sedimentation velocity @xmath9 of interacting colloidal particles depends both on the indirect hydrodynamic interactions ( hi ) mediated by the suspending solvent , and on the microstructure of the suspension . in equilibrium ,
the latter is determined by direct potential forces arising for example from the steric repulsion between the particles and from the electrostatic repulsion of overlapping double layers .
different pair potentials @xmath10 lead to rather different microstructures , and this in turn strongly effects the sedimentation .
it is well established by theories and experiments that the sedimentation velocity of a dilute suspension of monodisperse hard spheres is given by @xmath11 where @xmath1 is the particle volume fraction , and @xmath12 is the sedimentation velocity at infinite dilution @xcite . on the other hand , the long - ranged electrostatic repulsion occurring in suspensions of charged particles can give rise to a reduction in @xmath9 , as compared to a hard sphere dispersion at the same volume fraction @xcite . this decrease in @xmath9
is mainly due to the cumulative backflow of displaced fluid , which becomes particularly effective because the probability of two or more charged particles coming close to each other is very small .
conversely , hard spheres are effectively attracted to each other at small interparticle distances @xmath13 , where @xmath14 is the radius of the spheres .
this can be readily seen from the potential of mean force @xmath15 , where @xmath16 is the radial distribution function of hard spheres , which shows its maximum value at contact distance @xmath17 due to excluded volume effects .
closely spaced particles are mutually exposed to the downflow of nearby fluid , dragged along with the sedimenting particles . consequently , the retardation from backflow is reduced , whereas the influence of near - field hi is enhanced for hard sphere suspensions .
while the reduction in @xmath9 due to long - ranged electrostatic forces is known since many years , it was not realized until recently that , in particular for deionized charge - stabilized suspensions , @xmath9 can be significantly smaller than @xmath12 even for extremely small volume fractions as @xmath18 .
in fact , in the past the effects of hi have been frequently considered to be negligible for dilute suspensions of charged particles @xcite .
however , recent calculations have clearly demonstrated for these systems that hi is of importance at small @xmath1 not only for sedimentation @xcite , but also for short - time @xcite and long - time @xcite collective diffusion , and for long - time self - diffusion @xcite .
moreover , it was shown theoretically for dilute suspensions of charged colloids without added electrolyte that @xmath9 follows a non - linear @xmath1-dependence of the parametric form @xcite @xmath19 the numerically calculated coefficient @xmath20 was found to be nearly independent of the ( effective ) particle charge @xmath5 , provided that @xmath5 is kept large enough to completely mask the hard core of the particles .
experimenal results of sedimentation experiments on charged colloids agree favorable with the scaling - prediction of eq .
( [ u.charged ] ) @xcite .
similar non - linear volume fraction dependencies are found for the short - time translational and rotational diffusion coefficients of charged particles @xcite .
the same @xmath1-dependence as found for salt - free fluid suspensions of charged colloids is known to be valid for the sedimentation velocity of dilute ordered arrays of fixed spheres @xcite . for such arrays , the coefficient @xmath21 is determined analytically as @xmath22 for a @xmath23 lattice , and as @xmath24 for a @xmath25 or a @xmath26 lattice @xcite . as discussed in detail in refs .
@xcite , the main peak position @xmath27 of the radial distribution function @xmath16 for highly charged particles in salt - free suspension scales as @xmath28 which is also typical of a crystalline solid .
this , in fact , turns out to be relevant for eq .
( [ u.charged ] ) to be valid for strongly repelling particles like charged colloids and for ordered arrays of fixed spheres @xcite . in the present article ,
we analyse the sedimentation velocity of charged colloidal dispersions at low salinity both as a function of @xmath1 and of @xmath5 .
our numerical results for @xmath0 with system parameters representing monodisperse modified pmma particles investigated very recently in sedimentation experiments @xcite , are well described by eq .
( [ u.charged ] ) if the effective macroion charge is chosen such that @xmath29 .
we physically explain the scaling relation eq .
( [ u.charged ] ) in terms of an effective hard sphere ( ehs ) model by using wertheims analytical expression @xcite for the static structure function of hard sphere dispersions obtained in percus - yevick approximation . in this specific form of the ehs model , also the numerically determined value @xmath4
is recovered very accurately .
we will further demonstrate for weakly charged spheres that @xmath30 displays a volume fraction dependence proportional to @xmath31 , provided the suspension is completely deionized and @xmath1 is chosen small enough that the radial distribution function of the macroions can be approximated by its zero - density limit .
this result , however , turns out to be valid only for extremely small values of @xmath1 and @xmath5 , which we believe are not accessible in sedimentation experiments .
we start this section by summarizing the theoretical method used to calculate the reduced sedimentation velocity @xmath8 of charged colloidal spheres .
our results for @xmath8 are based on the effective macroion fluid model of charge - stabilized suspensions @xcite . in this model ,
the effective pair potential @xmath10 between two charged colloidal particles consists of a hard - core part with radius @xmath14 , and of a longer - ranged screened coulomb potential @xmath32 for @xmath33 , i.e. @xmath34 here , @xmath35 is a dimensionless coupling parameter given by @xmath36 where @xmath37 is the so - called bjerrum length , and @xmath38 is the thermal energy . the suspending fluid is treated as a continuum without internal structure , only characterized by its dielectric constant @xmath39 .
moreover , @xmath5 is the effective charge of a colloidal particle in units of the elementary charge @xmath40 .
the screening parameter @xmath41 is given by @xmath42=\kappa_c^2+\kappa_s^2,\ ] ] where @xmath43 is the number density of added 1 - 1-electrolyte , and @xmath44 is the number density of colloidal particles .
notice that @xmath41 comprises a contribution @xmath45 due to counterions , which are assumed to be monovalent , and a second contribution @xmath46 arising from added electrolyte .
( [ yukawa.potential ] ) is a good approximation for @xmath10 even for strongly charged colloids and for values of @xmath47 significantly larger than one , provided the effective charge number @xmath5 is regarded as an adjustable parameter @xcite . on the time
scales probed by dynamic light scattering and by sedimentation experiments , the effect of hi on the translational motion of the colloidal particles is embodied in the hydrodynamic diffusivity tensors @xmath48 ; @xmath49 @xcite .
the short - hand notation @xmath50 denotes the configuration of the @xmath51 spherical particles . without hi , @xmath52 , where @xmath53 is the stokesian diffusion coefficient of a particle with radius @xmath14 in a solvent with viscosity @xmath54 , and @xmath55 denotes the unit tensor .
the reduced short - time sedimentation velocity @xmath8 of a macroscopically homogeneous suspension of monodisperse colloidal spheres is then given by the zero - wavenumber limit @xmath56 of the so - called hydrodynamic function @xmath57 @xcite .
this function is defined as @xmath58 with wavevector @xmath59 of magnitude @xmath60 and corresponding unit vector @xmath61 .
the brackets indicate an equilibrium ensemble average .
@xmath57 can be regarded as a generalized ( short - time ) sedimentation coefficient of particles exposed to spatially periodic external forces aligned with @xmath62 , and derived from a weak potential proportional to @xmath63 $ ] @xcite . in principle
, one needs to distinguish the short - time sedimentation velocity @xmath9 , defined through eq .
( [ u.lim.h ] ) , from the long - time sedimentation velocity , which is determined in conventional sedimentation experiments . in dilute suspensions
, however , when the hi are well described as a sum of pairwise additive interactions , both quantities are identical , since then sedimentation does not perturb the equilibrium microstructure @xcite . at larger volume fractions , the microstructure becomes distorted from its equilibrium form , since @xmath64-body hi with @xmath65 becomes important .
this distortion leads to an additional change in the sedimentation velocity .
nevertheless , simulations of hard sphere suspensions show that the differences between the short - time and the long - time sedimentation velocities due to memory effects are rather small @xcite .
the assumption of pairwise additive hi is justified for the important case of dilute ( typically @xmath66 ) charge - stabilized suspensions at sufficiently low ionic strength , as considered here . for such systems ,
the particles are kept apart from each other due to their strong electrostatic repulsion .
then the @xmath51-body diffusivity tensors @xmath48 are well approximated by the two - body tensors @xmath67 + ( 1-\delta_{ij}){\mbox{\boldmath$b$}}({\mbox{\boldmath$r$}}_i-{\mbox{\boldmath$r$}}_j),\ ] ] where the term @xmath68 is excluded from the sum .
this approximation has been verified in the case of translational and rotational self - diffusion by considering also the leading three - body contribution to hi @xcite , and in case of the sedimentation velocity by comparison with a more elaborate method , known as the lowest order form of the @xmath69-expansion @xcite .
the two - body translational mobility tensors @xmath70 and @xmath71 are calculated by means of series expansions in powers of @xmath72 @xcite .
we only quote the leading terms for further reference @xmath73 + \frac{1}{2}\left(\frac{a}{r}\right)^3 \left[{\mbox{\boldmath$1$}}-3\hat{{\mbox{\boldmath$r$}}}\hat{{\mbox{\boldmath$r$}}}\right ] + { \cal o}(r^{-7}),\end{aligned}\ ] ] with @xmath74 . using eqs .
( [ hq],[dtt.pairwise ] ) , @xmath57 is expressed in terms of an integral @xmath75,\ ] ] involving these mobility tensors together with the radial distribution function @xmath16 . in this work ,
we use eqs .
( [ u.lim.h],[h.pairlevel ] ) together with the series expansions of @xmath70 and @xmath71 for calculating @xmath8 , by including contributions up to order @xmath76 .
when terms only up to @xmath77 in the series expansions of @xmath70 and @xmath71 are employed , @xmath8 is given explicitly as @xmath78,\ ] ] with @xmath79 .
for charge - stabilized suspensions , it is not necessary to account for many terms in the expansion of the two - body mobility tensors , since the integrals in eq .
( [ h.pairlevel ] ) converge rapidly because @xmath16 is essentially zero at small interparticle distances .
in contrast , many terms are needed for hard spheres .
for example , using the zero - density form @xmath80 for the radial distribution function of hard spheres , we obtain from eq .
( [ u.linear.phi ] ) that @xmath81 with @xmath82 . here , @xmath83 is the unit step function . by including terms only up to @xmath84 ,
the result is @xmath85 , as can be seen from eq .
( [ u.linear.phi ] ) , when the last term on the right hand side , arising from the term proportional to @xmath86 in eq .
( [ series.a ] ) is omitted . on the other hand ,
if terms up to @xmath87 are considered , the result for @xmath21 is improved to @xmath88 , which is close to the exact value @xmath89 , first obtained by batchelor using tabulated numerical results for the near - field hi @xcite . to obtain @xmath8 for
dilute charge - stabilized suspensions , we calculate @xmath16 in the effective macroion fluid model by using , for simplicity , the well - established rescaled mean spherical approximation ( rmsa)@xcite . on the basis of the pairwise additivity approximation of the hi combined with the @xmath72-expansion of the mobility tensors , henceforth referred to as pa - scheme , @xmath8 can then be calculated as a function of @xmath1 .
we show in the following that the exponent @xmath90 in eq .
( [ u.charged ] ) and the charge - independence of the parameter @xmath21 , both found from numerical calculations , can be understood _ quantitatively _ in terms of a model of effective hard spheres ( ehs ) , which can be treated _
analytically_. the ehs model accounts for the most important feature of the radial distribution function @xmath16 of highly charged particles , namely the so - called correlation hole .
for an illustration of that , consider fig .
[ gr.rmsa.ehs.plot ] , which shows a typical @xmath16 for a salt - free suspension of strongly charged particles at @xmath91 . due to the strong electrostatic repulsion between the particles
, @xmath16 has a well developed first maximum , and a spherical region with nearly zero probability of finding another particle .
this region is referred to as the correlation hole . for small @xmath1
, the correlation hole usually extends over several particle diameters @xcite .
therefore , we can approximate the actual @xmath16 of the charge - stabilized system by the radial distribution function of an effective hard sphere ( ehs ) system with an effective radius @xmath92 and an effective volume fraction @xmath93 .
the ehs - radius @xmath94 accounts for the electrostatic repulsion between the particles and can be identified as @xmath95 , where @xmath27 is the principal peak position of the actual @xmath16 . since the extension of the correlation hole is substantially larger than @xmath14 for small volume fractions and large particle charges , it is then a good approximation to consider only the leading oseen - term in @xmath71 ( cf .
( [ series.a],[series.b],[h.pairlevel ] ) ) in calculating the sedimentation velocity .
@xmath0 is then well approximated by @xmath96 here , @xmath97 is the total correlation function and @xmath98 denotes the laplace transform of @xmath99 .
next , we approximate @xmath100 by the total correlation function @xmath101 of the ehs model , evaluated at the effective volume fraction @xmath102 . with this approximation for @xmath100 used in eq .
( [ u.ehs ] ) , we readily obtain the result @xmath103 with a fractional exponent @xmath90 , and the parameter @xmath21 given by @xmath104 where @xmath105 is the laplace transform of @xmath106 with @xmath107 and @xmath108 .
we notice that @xmath102 and hence @xmath21 are indeed independent of @xmath1 and @xmath109 , provided that @xmath94 is identified with @xmath110 .
this follows from the fact that , for deionized suspensions of strongly charged particles where @xmath111 holds , @xmath27 coincides within @xmath112 with the average geometrical distance @xmath113 of two particles . for an illustration of this fact consider the inset of fig .
[ gr.rmsa.ehs.plot ] , which shows rmsa results for @xmath114 as a function of @xmath1 .
thus , the scaling relation @xmath115 holds and this gives rise to the exponent @xmath90 in eq .
( [ u.charged ] ) : by identifying @xmath94 with @xmath116 , we obtain an effective volume fraction @xmath117 independent of @xmath5 and @xmath1 . to obtain a numerical value of @xmath21 according to eq .
( [ a.ehs ] ) , we take now advantage of an analytic expression for @xmath105 given in the percus - yevick ( py ) approximation @xcite . by performing the zero-@xmath118 limit ,
we obtain in py approximation after a straightforward calculation the intermediate result @xmath119 substitution of eq .
( [ htilde.py ] ) into eq .
( [ a.ehs ] ) gives @xmath120 since @xmath121 , we obtain from eq .
( [ a.py ] ) a value @xmath122 remarkably close to @xmath123 as determined from the parametric fit of our numerical pa - result for @xmath0 ( cf . following section ) . for an illustration of the replacement of the actual @xmath16 by the ehs-@xmath16 within the ehs model , see again fig .
[ gr.rmsa.ehs.plot ] , which shows besides a typical rmsa-@xmath16 of charge - stabilized particles the corresponding ehs-@xmath16 obtained in py approximation for @xmath121 .
we finish this section with two remarks .
first , we would like to stress again the close connection between the relation eq .
( [ u.charged ] ) , found for highly charged colloidal suspensions , and the corresponding result for ordered arrays of fixed spheres .
for both systems , the @xmath124-scaling behavior of @xmath0 is caused by a strong structural correlation of the particles , resulting in the scaling relation @xmath28 @xcite .
although saffmann already mentioned the possibility for finding a scaling behavior as eq .
( [ u.charged ] ) for highly correlated fluid suspensions @xcite , the numerical method descibed here provides the first quantitative results on the sedimentation velocity of charged suspensions , leading to the predicted @xmath124-scaling in eq .
( [ u.charged ] ) plus the calculation of the prefactor @xmath21 .
it is further important to notice that the scaling relation @xmath125 is not valid if significantly large amounts of excess electrolyte are added to the suspension , since then the particle diameter becomes another physically relevant length scale besides the mean particle distance @xmath126 .
the @xmath1-dependence of @xmath30 at small @xmath1 changes then gradually with increasing @xmath43 from a @xmath124-dependence to the linear @xmath1-dependence of eq .
( [ u.hs ] ) characteristic for hard spheres @xcite .
it is also of interest to consider the opposite limiting case of dilute suspensions of weakly charged spheres in the framework of the effective macroion fluid model .
this case was addressed first by petsev and denkov @xcite .
suppose @xmath5 is so small that @xmath127 i.e. the electrostatic repulsion can be treated as a small perturbation of the hard - core repulsion between the particles .
for very small @xmath1 , @xmath16 can then be approximated as @xmath128\ ] ] with @xmath80 .
in other words , @xmath16 is approximated by its zero - density limit @xmath129 , linearized with respect to @xmath130 .
substitution of eq .
( [ gr.linearized ] ) in eqs .
( [ u.lim.h],[h.pairlevel ] ) gives two additive contributions to @xmath8 , the first one arising from the hard - core part @xmath131 , and the second one from the electrostatic perturbation of @xmath16 in eq .
( [ gr.linearized ] ) .
explicitly @xcite @xmath132,\ ] ] where the first term in the bracket is the result of eq .
( [ u.hs ] ) , arising from the hard - core part of @xmath16 .
as far as the electrostatic part is concerned , it was argued by petsev and denkov @xcite that the leading oseen - term in @xmath71 should give the dominant contribution to @xmath8 .
therefore only the oseen - term is considered here , leading to the second term in the bracket of eq .
( [ u.petsev ] ) .
we will provide a critical discussion of this approximation in the following section .
next , we further simplify eq .
( [ u.petsev ] ) for the case of vanishing excess electrolyte ( @xmath133 ) , where @xmath134 with @xmath135 . when @xmath1 and/or @xmath136 are sufficiently small so that @xmath137 , eq .
( [ u.petsev ] ) is further simplified to @xmath138^{\frac{1}{2 } } \phi^{\frac{1}{2}}+\cal{o}(\phi).\ ] ] thus , we expect the reduced sedimentation velocity of dilute deionized suspensions of weakly charged particles to scale like @xmath139 , with a parameter @xmath21 depending on the macroion charge @xmath5 and on the ratio @xmath140 .
notice that the hard core contribution @xmath141 to @xmath8 in eq .
( [ u.petsev ] ) is omitted against the term proportional @xmath31 in proceeding from eq .
( [ u.petsev ] ) to eq .
( [ u.phi.sqrt ] ) . therefore , eqs .
( [ u.petsev ] ) and ( [ u.phi.sqrt ] ) do not become equal to each other for @xmath142 . whereas eq .
( [ u.petsev ] ) reduces to the hard sphere result of eq .
( [ u.hs ] ) for @xmath143 , eq .
( [ u.phi.sqrt ] ) results in @xmath144 for vanishing particle charge . obviously , the range of validity of eq .
( [ u.phi.sqrt ] ) is restricted to small values of @xmath5 , but still sufficiently large that the second term in eq .
( [ u.petsev ] ) plays the dominant role compared to the hard core contribution @xmath141 .
we mention already here that due to the strong approximations made in deriving eqs .
( [ u.petsev],[u.phi.sqrt ] ) , the range of validity of these two equations is restricted to extremely small values of @xmath1 and @xmath5 ( cf . following section ) . for eqs .
( [ u.petsev],[u.phi.sqrt ] ) to be valid , it is further assumed that the van der waals attraction among the particles is negligibly small .
experimentally , this might be achieved by coating the particle surface with a thin layer of polymer chains and/or by matching the solvent refractive index to the refractive index of the colloidal particles
. if attractions between the particles are present , not only eqs .
( [ u.petsev],[u.phi.sqrt ] ) cease to be valid , but also sedimentation velocities larger than for hard spheres may be measured at the same density @xcite .
this enhanced sedimentation rate is due to the enhanced probability , as compared to hard spheres , of two particles for being close together , thus leading to a reduced retardation from backflow @xcite .
in this section , we present numerical results for the reduced sedimentation velocity @xmath8 as a function of the volume fraction @xmath1 and of the particle charge number @xmath5 .
our pa - scheme calculations of @xmath0 account , if not stated differently , for two - body contributions to the hydrodynamic mobility tensors up to order @xmath76 .
the system parameters employed in our calculations are @xmath145 ( corresponding to cis - decalin as dispersing fluid at temperature @xmath146k ) , and particle radius @xmath147 nm , representing modified pmma particles investigated very recently in sedimentation experiments @xcite . we further use @xmath133 , i.e. the ionic strength is essentially determined by the ( monovalent ) counterions , which counterbalance the charge of the colloidal particles .
obviously , the dielectric constant @xmath39 used in the calculations presented here is rather small , leading to strong long - ranged repulsions between the particles even for small surface charges , which , most probably , are present in any polar organic solvent @xcite .
furthermore , for these solvents , residual water can cause strong electric charging of the pmma particles @xcite .
we therefore present calculations for both weakly and strongly charged particles .
our calculations for strongly charged particles recover the results obtained in refs .
@xcite for different system parameters , demonstrating clearly the independence of the scaling relation eq .
( [ u.charged ] ) from specific system parameters , in particular from the solvent dielectric constant @xmath39 . in fig .
[ u.highcharge.plot ] , we show the pa - scheme result for @xmath8 , obtained by choosing an effective charge number @xmath148 large enough that the physical hard - core radius @xmath14 of the particles constitutes no relevant physical length scale .
this numerical result is perfectly fitted by the form @xmath149 , with @xmath123 and @xmath150 . as mentioned before , this is in remarkably good _ quantitative _ agreement with our ehs model result obtained in py approximation . as discussed in detail in ref .
@xcite , adding small amounts of excess electrolyte leads to a significant increase in @xmath9 , and the @xmath124-scaling behaviour of @xmath8 does not hold any more . for comparison ,
[ u.highcharge.plot ] includes also the reduced sedimentation velocity of uncharged hard spheres according to eq .
( [ u.hs ] ) . evidently , the sedimentation velocity of charged particles decreases much faster with increasing @xmath1 than one would expect for hard spheres at the same volume fraction . as shown in fig .
[ gr.rmsa.ehs.plot ] , the rmsa-@xmath16 corresponding to the largest volume fraction @xmath91 considered in fig .
[ u.highcharge.plot ] has well developed undulations , with the maximum value approximately located at the mean geometrical particle distance @xmath116 .
we only quote that the corresponding static structure factor @xmath151 has its principal peak height well below the range of values @xmath152 , where the system starts to freeze according to the empirical hansen - verlet rule @xcite .
a few comments on the quality of the rmsa input for @xmath16 , used in the pa - scheme calculations , is in order here .
it is well known @xcite that the rmsa underestimates to some extend the oscillations of @xmath16 and @xmath151 in case of strongly correlated systems of highly charged particles . in principle , we could use instead of the rmsa an alternative scheme like the rogers - young ( ry ) scheme @xcite .
the ry sheme is quite accurate within the effective macrofluid model , but has the disadvantage of being numerically far more involved than the rmsa .
fortunately , the rmsa has been found to give nearly identical results as the ry - scheme , provided that a somewhat larger value of @xmath5 is used in the rmsa calculations @xcite . we have argued before and will show in the following that the parameter @xmath21 in eq .
( [ u.charged ] ) is nearly independent of @xmath5 , typically as long as @xmath153 .
consequently , using the ry - scheme instead of the rmsa for the same value of @xmath5 should lead to practically identical results for @xmath8 .
we have verified this assertion by explicit ry calculations @xcite . in fig .
[ u.z.plot ] , we display pa - scheme results for @xmath8 versus @xmath1 for various values of @xmath5 . for @xmath143
, we recover the hard sphere result @xmath81 with @xmath88 .
an increase in @xmath5 leads to a strong reduction in @xmath9 , with a gradual transition from the linear @xmath1-dependence of @xmath8 towards the non - linear form of eq .
( [ u.charged ] ) , with @xmath154 .
this figure nicely illustrates that @xmath8 becomes independent of @xmath5 for @xmath153 . for these large values of @xmath5
, @xmath155 stays practically constant when @xmath5 is increased with fixed @xmath1 @xcite .
the effect on the pa - results for @xmath8 caused by truncating the @xmath72-expansion of the two - particle mobilities after various terms of increasing order in @xmath72 , can be assessed from fig .
[ u.divcontrib.plot ] .
the solid line represents the full pa - scheme result where all two - body contributions up to @xmath87 are accounted for .
nearly identical results for @xmath0 are obtained , even at @xmath91 , when contributions only up to @xmath77 are considered .
in contrast , the result for @xmath0 obtained by accounting only for hydrodynamic contributions up to @xmath84 shows clear deviations from the full pa - scheme result at volume fractions @xmath156 . as a conclusion
, we can state that for dispersions of highly charged particles it is justified to use a truncated far - field expansion of the mobility tensors with only the first terms being included , provided a good static input for @xmath16 is used .
the finding that an expansion up to @xmath77 leads already to good results indicates further that hydrodynamic @xmath64-body contributions to @xmath9 with @xmath65 are indeed negligibly small in the considered @xmath1-range . for practical purposes , it is therefore suitable to use the simple eq .
( [ u.linear.phi ] ) for calculating the reduced sedimentation velocity for highly charged suspensions .
we discuss now the behaviour of @xmath8 when @xmath5 is small . to be specific , consider first the value @xmath157 .
for this charge number , @xmath158 .
this implies that the microstructure is affected not only by the electrostatic repulsion but also by the physical hard core of the particles . in fig .
[ u.div.gr.plot ] , we have redrawn from fig . [ u.z.plot ] the pa - scheme result for @xmath8 with @xmath157 and rmsa - input for @xmath16 .
this graph should be compared with the corresponding result for @xmath8 obtained from the expression given in eq .
( [ u.petsev ] ) ( henceforth referred to as pd - result ) .
the pd - result for @xmath8 is completely different from the corresponding pa - scheme result , in particular at larger @xmath1 where the pd-@xmath8 even turns negative .
the failure of the pd - expression in describing @xmath0 arises from the fact that the approximations entering in its derivation , i.e. in particular the approximation eq .
( [ gr.linearized ] ) for @xmath16 , but also the truncation of the electrostatic contribution to @xmath8 after the leading oseen - term , are not justified even for a charge number as small as @xmath157 .
the consideration of more terms in the hydrodynamic pair mobility contributions up to @xmath87 , however , leads only to a small increase in @xmath8 , as can be seen from fig .
[ u.div.gr.plot ] , which further includes the pa - scheme result for @xmath0 with @xmath16 approximated by the linearized zero - density form of @xmath16 , eq .
( [ gr.linearized ] ) .
another slight improvement in @xmath8 is achieved when the non - linearized zero - density form @xmath159 is used as input in the pa - scheme .
in fact , the contact value @xmath160 is not small enough for @xmath157 to fully justify a linearization in @xmath130 .
therefore , the main reason for the failure of the pd - expression is due to the significant differences of the actual @xmath16 at @xmath157 from its zero - density form already for @xmath161 . to obtain further insight in the range of validity of the pd - result eq .
( [ u.petsev ] ) , we display in fig .
[ u.zsmall.plot ] results for @xmath8 for an even smaller particle charge @xmath162 and volume fractions smaller than @xmath163 . as seen from this figure and from fig .
[ u.div.gr.plot ] , the differences between the rmsa - pa - scheme result for @xmath0 and the pa - scheme result for using @xmath164 or @xmath165 $ ] as static input become smaller with decreasing @xmath5 , but are still significant even for @xmath162 and @xmath166 . for an explanation of this finding , we refer to fig .
[ gr.z2.plot ] , where the radial distribution @xmath16 obtained from the rmsa and from the zero - density expression @xmath164 for @xmath162 and @xmath167 are shown . obviously , there are still small differences between the the zero - density expression for @xmath16 and the corresponding rmsa result even for these small values of @xmath5 and @xmath1 .
we show now that these small differences in @xmath16 cause the large differences in the results for @xmath8 as illustrated in fig .
[ u.zsmall.plot ] . for this purpose ,
we plot in the inset of fig .
[ gr.z2.plot ] the function @xmath168 as obtained from the radial distribution functions shown in the main figure .
as easily seen , the two curves for @xmath168 , corresponding to the two different inputs for @xmath16 , are remarkably different from each other even up to large distances @xmath169 .
it is now crucial to notice that according to eq .
( [ u.linear.phi ] ) it is essentially the function @xmath168 and not @xmath16 itself which appears in the integrand when @xmath8 is calculated including terms up to @xmath170 in the series expansions in eqs .
( [ series.a],[series.b ] ) .
therefore , the small differences in the two @xmath16 s considered in fig .
[ gr.z2.plot ] give rise to substancial differences in the corresponding sedimentation velocities shown in fig .
[ u.zsmall.plot ] .
we can therefore conclude that it is absolutely important to employ an accurate @xmath16-input for calculating @xmath8 in the pa - scheme . due to the use of the ( linearized ) zero - density approximation of @xmath16 , the expression given by petsev and denkov in eq .
( [ u.petsev ] ) does not predict the sedimentation velocity of weakly charged particles correctly .
after having explored the high sensibility of @xmath8 on the form of the @xmath16-input used in the pa - scheme , it is now apparent that the second approximation made in deriving eq .
( [ u.petsev ] ) , i.e. the omission of terms in the far - field expansions of the hydrodynamic mobilities of @xmath84 in the electrostatic contribution to @xmath8 , is not valid for the small values of @xmath5 and @xmath1 used in figs .
[ u.zsmall.plot ] and [ gr.z2.plot ] .
this conclusion follows from fig .
[ u.zsmall.plot ] , when the results for @xmath0 according to eq .
( [ u.petsev ] ) and derived from the pa - scheme using eq .
( [ gr.linearized ] ) as static input are compared .
although the same approximation is employed for @xmath16 , the two results for @xmath0 do not agree , since the approximations in truncating the series expansions of the hydrodynamic mobility tensors after the terms of @xmath170 and @xmath87 , respectively , are not the same .
on the other hand , we have shown in fig .
[ u.divcontrib.plot ] for highly charged particles that the first few terms in the series expansion of eqs .
( [ series.a],[series.b ] ) are sufficient to calculate @xmath8 . as pointed out before , this is due to the existence of an extended correlation hole in the @xmath16 of highly charged particles in deionized suspensions ( cf .
[ gr.rmsa.ehs.plot ] ) .
the correlation hole gives rise to a fast concergence of the integrals in eq .
( [ h.pairlevel ] ) .
however , in case of the weakly charged particles considered here , there is no correlation hole present , as can be seen , e.g. , from fig .
[ gr.z2.plot ] .
therefore , a large number of terms in the series expansions of eqs . ( [ series.a],[series.b ] ) are needed for weakly charged particles to calculate @xmath0 .
in fact , as discussed before in the case of hard spheres , the terms of @xmath84 and @xmath77 in the series expansions in eqs .
( [ series.a],[series.b ] ) contribute in approximately the same weight to @xmath8 , still remarkably large compared the the leading oseen - term of @xmath170 , considered only in the electrostatic contribution in the pd - result eq .
( [ u.petsev ] ) . to summarize the last presented results
, we have shown that the range of validity of eq .
( [ u.petsev ] ) is restricted to values of @xmath1 considerably smaller than @xmath171 , where the macroion radial distribution function is described very accurately by its linearized zero - density form . furthermore , for the pd - result eq .
( [ u.petsev ] ) to be hold , the effective charge @xmath5 has to be small enough that the approximation in eq .
( [ gr.linearized ] ) can be used to calculate @xmath8 , but still not so small that the actual @xmath16 is too close to its zero - charge limiting form @xmath80 . otherwise , the omission of higher order hydrodynamic terms of @xmath84 to the electrostatic contribution in eq .
( [ u.petsev ] ) ceases to be a good approximation .
these severe restrictions limit the range of validity of eq .
( [ u.petsev ] ) to values of the system parameters most probably not accessible in sedimentation experiments .
moreover , the relative differences in @xmath9 and @xmath172 become small in the parameter range where eq .
( [ u.petsev ] ) should apply .
let us now turn to some remarks on the @xmath31-scaling of @xmath30 proposed in eq .
( [ u.phi.sqrt ] ) .
( [ u.phi.sqrt ] ) is derived from the pd - result eq .
( [ u.petsev ] ) , the before discussed restricted range of validity of eq .
( [ u.petsev ] ) applies also to eq .
( [ u.phi.sqrt ] ) .
furthermore , in eq .
( [ u.phi.sqrt ] ) , the terms linear in @xmath1 are neglected against the term proportional to @xmath31 .
we have shown in fig .
[ u.zsmall.plot ] that this additional approximation becomes invalid already at small volume fractions , typically @xmath166 . for larger @xmath1 , the two results for @xmath0 in eq .
( [ u.petsev ] ) and eq .
( [ u.phi.sqrt ] ) start to deviate strongly , since the contributions to @xmath8 linear in @xmath1 ( especially the hard core contribution @xmath141 ) are no longer small compared to the term proportional to @xmath31 .
we therefore also expect the @xmath31-scaling of @xmath30 according to eq .
( [ u.phi.sqrt ] ) not to be measurable in a sedimentation experiment .
nevertheless , we wish to point out that there is a specific range of particle charges @xmath5 where @xmath30 indeed scales as @xmath31 for a broad range of volume fractions .
this range of @xmath5-values is determined when the effective particle charge is subsequently lowered from large values @xmath173 , where @xmath30 scales like @xmath124 , to @xmath143 , where @xmath30 behaves linearly in @xmath1 according to eq .
( [ u.hs ] ) . to demonstate the occurance of a @xmath31-dependence of @xmath30 , we have redrawn in fig .
[ u.phisqrt.plot ] the rmsa - pa result from fig .
[ u.div.gr.plot ] for @xmath157 together with the corresponding results for @xmath174 and @xmath175 .
obviously , all three curves are well described by the form @xmath176 with a parameter @xmath6 very close to @xmath7 .
such a @xmath31-dependence of @xmath8 might indeed be measurable in sedimentation experiments on dispersions with suitably chosen particle charges .
however , we stress that the occurance of the parameter @xmath6 close to @xmath7 for a broad range of volume fractions as dipicted in fig .
[ u.phisqrt.plot ] is completely different from the prediction eq .
( [ u.phi.sqrt ] ) , which has been shown to be valid only for very small @xmath1 . finally , we shortly comment on the dependence of our results on the system parameters @xmath177 and @xmath39 , which are held fixed in our calculations .
obviously , these two parameters enter into our calculations mainly by determining the bjerrum length @xmath178 , which has a rather large value @xmath179 nm due to the small dielectric constant @xmath145 used in this work . as consequence ,
also the ratio @xmath140 is comparatively large .
this ratio determines together with @xmath5 the strength of the electrostatic repulsion between the particles .
for that reason , one might argue that our dicussion concerning the use of eq .
( [ gr.linearized ] ) for @xmath16 in calculating @xmath8 is not longer valid for smaller values of @xmath178 .
however , explicit pa - scheme calculations for system parameters as used in ref . @xcite and/or
a simple estimate show that our conclusions derived above remain valid even for considerably smaller values of @xmath178 : consider a value of @xmath180 ten times smaller than the one used in the above presented calculations , e.g. a value like in sedimentation experiments on charged silica particles in ethanol @xcite .
then , according to eqs .
( [ yukawa.potential],[yukawa.coupling ] ) , one might choose a particle charge approximately @xmath181 times larger than in our calculations to achieve the same strength of repulsion between the particles .
as shown above for our choice of system parameters , the use of eq .
( [ gr.linearized ] ) for @xmath16 is a poor approximation even for low volume fractions @xmath182 and a particle charge @xmath162 .
therefore , even for a system with @xmath180 ten times larger , eqs .
( [ u.petsev],[u.phi.sqrt ] ) are applicable only for particle charges considerably smaller than @xmath174 . this slightly extended parameter range for eqs .
( [ u.petsev],[u.phi.sqrt ] ) to be valid is most probably still too restricted to be experimentally accessible .
we have presented theoretical results for the reduced sedimentation velocity @xmath8 of monodisperse charged suspensions in dependence of the volume fraction @xmath1 and of the particle charge number @xmath5 . our theoretical model for @xmath8
is based on the effective macroion fluid model and on the assumption that pairwise additive hi prevails at sufficiently small @xmath1 .
the numerical results for @xmath8 in case of strongly charged particles at low salinity are well parametrized by the form @xmath183 .
we have shown that the exponent @xmath90 and the value of the charge - independent parameter @xmath184 can be _ quantitatively _ understood in terms of a model of effective hard spheres of radius @xmath94 which depends on the volume fraction .
using percus - yevick input for the static pair correlation function of hard spheres , the ehs model can be treated analytically and leads to a value of @xmath21 very close to @xmath185 .
it was further demonstrated that @xmath30 can scale like @xmath31 in case of dilute suspensions of very weakly charged particles .
this peculiar volume fraction dependence of @xmath8 derives from an expression given by petsev and denkov @xcite , when it is further assumed that the ionic strength in the system is mainly due to counterions .
our numerical calculations reveal that both the original expression for @xmath8 given by petsev and denkov , and the derived expression showing the @xmath31-scaling of @xmath30 , are only valid for very small particle volume fractions , which we expect not to be accessible in conventional sedimentation experiments .
we are grateful to barbara lhle ( university of konstanz ) for providing rogers - young calculations of various static distribution functions , and to bruce ackerson ( oklahoma state university ) and barbara mandl ( formerly university of konstanz ) for helpful discussions .
we further thank on of the referees for calling our attention to ref .
thanks the deutsche forschungsgemeinschaft for financial support within sfb 513 and sfb 237 . 10 , , 245 ( 1972 ) . and b. u. felderhof , , 213 ( 1989 ) . and p. p. j. m. schram , , 3137 ( 1992 ) . and h. m. fijnaut , , 1618 ( 1981 ) . , j. w. goodwin , r. h. ottewill , and t. f. tadros , , 78 ( 1982 ) . , a. p. philipse , and h. n. w. lekkerkerker , , 427 ( 1996 ) . , a. nikolov , and d. t. wasan , , 160 ( 1998 ) . and
j. l. anderson , in _ hydrosols and rheology _ , edited by m. kerker , volume 4 of _ colloid and interface science _ , academic press , new york , 1976 . and j. l. anderson , , 816 ( 1980 ) . and w. van megen , , 194 ( 1984 ) . , , 578 ( 1980 ) . and
r. j. a. tough , in _ dynamic light scattering _ , edited by r. pecora ,
plenum , new york , 1985 .
, in _ liquids , freezing and glass transition : ii _ , edited by j .-
hansen , d. levesque , and j. zinn - justin , north holland , amsterdam , 1991 . , b. steininger , u. genz , and r. klein , , 119 ( 1994 ) . , , 215 ( 1996 ) . ,
a. p. philipse , g. ngele , b. mandl , and r. klein , , 43 ( 1995 ) . , o. kellerbauer , r. krause , and r. klein , , 2562 ( 1993 ) . , b. mandl , and r. klein , , 117 ( 1995 ) . , g. ngele , and r. klein , , 6224 ( 1996 ) . and p. baur , , 297 ( 1997 ) . , oklahoma state university , private communication . and g. ngele , , 56 ( 1997 ) . and g. ngele , , 168 ( 1997 ) . and g. ngele , , 1258 ( 1997 ) . , , 317 ( 1959 )
, , 115 ( 1973 ) . and g. m. homsy , , 13 ( 1982 ) . ,
r. j. phillips , j. c. lester , and g. bossis , , 257 ( 1988 ) .
, , 321 ( 1963 ) . and p. n. pusey , , 137 ( 1991 ) . , , volume 2 of _ studies in interface science ( eds .
d. mbius and r. miller ) _ , elsevier , amsterdam , 1996 . and a. b. glendinning , , 948 ( 1981 ) . ,
d. a. saville , and w. r. schowalter , , cambridge university press , cambridge , 1989 . and
w. b. russel , , 124 ( 1982 ) .
, , 3484 ( 1990 ) . , ,
299 ( 1993 ) . and p. mazur , , 349 ( 1984 ) . and r. klein , , 26 ( 1991 ) . and r. schmitz , , 373 ( 1988 ) . , b. u. felderhof , and r. schmitz , , 383 ( 1988 ) . and n. d. denkov , , 329 ( 1992 ) . and d. n. petsev , , 462 ( 1992 ) . and h. k. schachman , , 19 ( 1955 ) . , c. g. de kruif , and a. vrij , , 501 ( 1986 ) .
, , 200 ( 1997 ) . and l. verlet , , 150 ( 1969 ) . and d. a. young , , 999 ( 1984 ) . ,
thesis , university of konstanz , 1996 .
, b. lhle , and g. ngele , unpublished results . | we report on calculations of the reduced sedimentation velocity @xmath0 in homogenous suspensions of strongly and weakly charged colloidal spheres as a function of particle volume fraction @xmath1 . for dilute suspensions of strongly charged spheres at low salinity ,
@xmath0 is well represented by the parametric form @xmath2 with a fractional exponent @xmath3 and a parameter @xmath4 , which is essentially independent from the macroion charge @xmath5 .
this non - linear volume fraction dependence can be _ quantitatively _ understood in terms of a model of effective hard spheres with @xmath1-dependent diameter .
for weakly charged spheres in a deionized solvent , we show that the exponent @xmath6 can be equal to @xmath7 , if an expression for @xmath8 given by petsev and denkov [ _ j .
colloid interface sci . _ * 149 * , 329 ( 1992 ) ] is employed .
we further show that the range of validity of this expression is limited to very small values of @xmath1 and @xmath5 , which are probably not accessible in sedimentation experiments .
the presented results might also hold for other systems like spherical proteins or ionic micelles .
* key words : * sedimentation , hydrodynamic interactions , charge - stabilized colloids * short running title : * sedimentation of charged particles |
web applications constitute valuable up - to - date tools in a number of different cases .
one such case is their use in the management of environmental problems so as to protect civilians from any unfortunate consequences that these problems can cause . their evolution , therefore , has been especially important in many cases , one of them being in the development of systems of administration of the air quality @xcite.the right of accessing environmental information has been enacted in european level through appropriate legislation , which are incorporated in the relevant greek legislation see @xcite- @xcite .
nowadays , the combination of telecommunications and new technologies create a framework for developing such systems increasingly sophisticated @xcite- @xcite .
that is just diffusion of environmental information and public access which was attempted effectively through the system codenamed eap ( laboratory of atmospheric pollution and environmental physics ) in western macedonia .
it is developed for the first time in 2002 @xcite , providing the possibility for direct information to the public about the air quality , as it was recorded in the four atmospheric measurement stations established in the capitals of countries kozani , florina , kastoria and grevena though an appropriate web - site , as well as sms , with the possibility for extension of stations and also the historical measurements privilege @xcite . for every station a previous and current index of pollution appears ( in a scale 1 - 10 ) with an appropriate colour scale @xcite .
the system was expanded and upgraded in may 2010 , which consists in transferring data , the way of presentation as well as the amount of information provided .
specifically it is recommended : a ) the combine use of different methods of transportation in real or almost real time data of terminal stations measurements to a central base station b ) the environmental information is promoted to the internet , with a properly designed dynamic website with enabled navigating of google map @xcite , @xcite , @xcite . in this paper
the novelty of information system eap is presented , in a dynamic , easily accessible and user - friendly manner .
it consists of a structured system that users have access to and they can manipulate thoroughly , as well as of a system for accessing and managing results of measurements in a direct and dynamic way .
it provides updates about the weather and pollution forecast for the next few days ( based on current day information ) in western macedonia .
these forecasts are displayed through dynamic - interactive web charts and the visual illustration of the atmospheric pollution of the region in a map using images and animation images .
moreover , there is the option to view historical elements .
an additional new function is the use of online reports to monitor , analyze , control and processing measurements , historical data and statistics of each station in real time over the internet .
this function focuses on designing an effective and user - friendly process . finally , the management system of measurement stations , the administrator has the ability to dynamically create , modify and delete objects , points and information of each station on the googlemap . in this way the processing ( update , delete ,
add ) of points is easier .
the a.q.m.e.i.s .
application has been developed using open source software tools like html , javascript , php and mysql .
html is the language for the internet interface design .
the goal for html was to create a platform - independent language for constructing hypertext documents to communicate multimedia information easily over the internet @xcite .
javascript is a client - side scripting language that provides powerful extensions to the html used to compose web pages and is mainly utilised for checking and validating web form input values to make sure that no invalid data are submitted to the server @xcite .
php is the most popular server - side programming language for use on web servers .
any php code in a requested file is executed by the php runtime , usually to create dynamic web page content .
it can also be used for command - line scripting and client - side gui applications .
php is a cross platform programming language and can be used with many relational database management systems ( rdbms)@xcite . mysql is a high - performance , multi - thread , multi - user rdbms and is built around a client - server architecture .
also mysql uses the structured query language standard giving a great deal of control over this relational database system @xcite .
finally , apache server is responsible for expecting requests from various programs - users and then serve the pages , according to the standards set by the protocol http ( hypertext transfer protocol ) @xcite .
in this section the user interface and functions of this application are described .
there are three ( 3 ) levels of user access ( groups of users ) . on the first level the user has the ability to be informed in real time about the weather conditions , the air pollution and air pollution indices in an area of interest using google map.the second level is for authorized users only , who can be informed analytically through reports about measurements of a specific time period .
the third level is for the administrator , who has access to all information and who also inserts , updates or deletes data from the database .
the administrator can also interfere dynamically and manage all the information of the googlemap .
the online web station reports is a new online web feature which offers to the approved members of the application to monitor , analyze , check and process the measurements using statistics of each station in real time .
furthermore , the ability of pumping previous measurements is given ; a function that did not exist in the web application until a short time ago .
the login is achieved through the use of a special personal password that is given to the members by the support group of the web application .
the users input the password and after validation , they can perform a number of available functions in a safe and user - friendly online web environment .
more specifically the feature offers the following functions to its members : presentation of daily , weekly , monthly and according to the user s choice values either for a chosen station of all its measured data or for specifically chosen measured parameters ( sensors ) , with simultaneous calculation and presentation of maximum , minimum , average values , sum , number and percentage of measurements in a table or the ability of output of data in ms excel .
the functions of the new web features are described in detail .
there are four categories of online web station reports , i.e. daily , weekly , monthly and custom .
each report is displayed in three parts ( forms ) . in the first form , by choosing a station
the image is displayed as well as various information about the specific station ( fig .
1 ) . if the user does not choose a station , an error report appears . by clicking on next , the second form of the report appears in which the user can choose which measure fields to be shown , as well as the measure time interval ( 5min or 60min ) and the specific date that those measurements were taken ( fig .
2 ) .
the dates differentiate according to the report category that the user will choose ; more specifically , there are : a ) daily report : the current date appears .
b ) weekly : the first day of the current week is set as the starting date .
c ) monthly : the first day of the current month is set as the starting date .
d ) periodical : the current dates are set as the dates ( from - to ) with the ability of changing the spaces ( from - to ) by the user if the user does not choose any measure fields or chooses date in which there are no figures reported , then the system will display an error message . by clicking on report the algorithm moves to the last tab of the report where a table of contents appears in a dynamic way with information about the hour , the measure fields , measurement results as well as other statistical data ( fig .
3 ) . also , the algorithm calculates and displays the number of measurements ( e.g. 100 measurements were found ) , the current page and the total number of pages ( e.g. page 1 from 12 ) .
depending on the number of records , an equal number of pages is created .
the application can display 25 measurements per page .
also the users can move to any page they wish so as to have access to any measurement of interest .
every form of measurement also has a status field ( numbers 0 , 1 or 2 ) .
this table is used to check the validity of the measurements of a field . in this way , if there are no results for a specific date in one field , then the indication no data is displayed .
all checks are made based on the status field .
if , however the measurements in a field are wrong for a specific date due to various factors , then the indication offscan is displayed . in a similar manner
, checks rely on the status field . for every measure field in a specific moment
the following statistics are taken into account a ) average , b ) minimum value , c ) date and time that minimum value was found , d ) the number of records of the minimum value , e ) maximum value , g ) the date and time of record of the maximum value , h ) the total number of measurements , i ) the % percentage . by pressing excel the measurements of the station can be displayed on excel form , which the user is able to open or save it for later use ( fig .
4 ) .
another innovation of the web application is the dynamic management of the measurements from each station in a simple way using an online geographical web interface .
the administrator using this specific feature ( management of the measurement stations using googlemaps ) can insert , delete and modify easily and simply data having as a purpose the dynamic renewal on the googlemap .
this gives the advantage to the administrator to use the specific feature as a platform of visualization of information without having to write on their own not even a line of code for this purpose .
moreover , an important element of the feature is the easy expansion and integration n ( n = count ) measurement stations on the interactive googlemap
. station management is made through the interactive interface of googlemap ; the administrator of this application can insert , delete and modify dynamically a certain point ( station ) in an area ( according to geographical latitude and longitude ) . to insert a certain station in the map ,
the following actions are required : a ) the insertion of municipality of choice : the user chooses through a list the one which the station belongs to ; b ) the insertion of the type of station of measurement : the user chooses if the station is meteorological or one that measures pollution or both .
all the data is stored in the application mysql database . as a last step
, the administrator sets the name of the station , the longitude and latitude , municipality , address , description , type of station and the image of the station .
( fig . 5 ) .
next , all information are stored into the database and are retrieved from there to be displayed dynamically ( both the points and information ) on the googlemap . on the map users
can see meteorological information as well as information about pollution from various stations and areas . for every station a previous and current index of pollution appears ( in a scale 1 - 10 ) with an appropriate corresponding colour scale . by clicking on each point of the station the information ( i.e. online measurements , air pollution indices for the previous and current day , general information about the station )
is displayed .
the user may also activate or deactivate one or more points on the map @xcite . to achieve the dynamic update of the measurement stations on the googlemap
, the file airlab_markers.php is called .
it is responsible for the creation and update of the xml file .
more specifically , the data of the application , i.e. the name and the measurements of the station , their geographical position where they belong , the general information with the representative photograph of each station , even the representation symbol , are retrieved in xml structure , submitting the appropriate preset sql question to the database , via the corresponding code of the php page .
the xml file has an element ( root - top level element ) and especially the <markers > <
/markers>. the remaining elements are nested to this . for the appropriate structure of the xml
file , there is an additional code in the airlab_markers.php file .
all necessary checks about the validity of the data then take place .
the algorithm was realized by php scripts and the specific feature that was developed , is supported by the internet explorer , firefox , opera and google chrome browsers .
an additional new feature is the weather forecast with the aid of dynamic web charts , is committed to deliver the most reliable , accurate weather information possible .
it provides free , real - time and online weather information for the web users with the state - of - the - art technology monitors conditions and forecasts in the area of western macedonia in the next few days ( fig .
the information is produced in a high - end server in eap / wmaqis @xcite and is read and stored in a database and it appears in the internet with the form of dynamic web graphs ( fig.7 ) .
meteorological parameters are temperature , humidity , wind speed , wind direction , accumulated precipitation , mixing height and total solar radiation .
php scripts retrieves from the mysql database the 24 average hourly values for each meteorological parameter ( except for the accumulated precipitation , for which a total of 6 hours is taken into account ) on each location in western macedonia .
next , the information is displayed with a graph .
the user then can choose a location to see the weather forecast . by choosing history ,
the previous meteorological measurements and figures are displayed using graphs .
another important part of the a.q.m.e.i.s .
application is the atmospheric pollution forecast of the pm10 ( particulate matter ) concerning the next few days in western macedonia .
our application displays dynamically these regions in a map using images ; according to pollution percentages in a certain region , the corresponding colour scale is represented denoting the levels of pollution .
choosing region , pollutant agent , source of emission and date , pollution for the previous and current dates , as well as the ones of the next three days are displayed .
this part of the application uses javascript , while a very small part of the code was written in php ( dates management ) .
the air pollution model produces image files ( xxx.jpg ) in the hard disc of the server in which javascript searches and then displays . in cases where not enough environmental data exist for a certain date ,
an image appears entitled pollution image display unavailable. the necessary validation integrity checks of the dates are also made ( from - to ) . finally choosing history the user can see older images of pollution rates in a certain region of western macedonia .
choosing movement , a javascript algorithm is executed animating the pollution illustrations in the area of interest .
( fig 9 ) .
mysql is a very fast and powerful database management system . allows the user to store , search , sort and recall data efficiently .
this application stores all information in the mysql database , so that they can be retrieved dynamically every time they are needed .
the architecture of this database consists of a total of 23 tables .
8 tables ( s001t05 , s001t60 , s002t05 , s002t60 , s003t05 , s003t60 , s004t05 , s004t60 ) include measurements collected from the stations ; they are used in reports and in the dynamic system for monitoring the air pollution , via an interactive chart .
s00x refers to station x and t05 or t60 refer to a measure time interval ( i.e. an average 5 or 60 minute measurement ) .
the primary key is date_time , while the rest of the fields ( value1 up to value32 ) store meteorological and environmental data .
tables city_info , points_categories and points are used for the management of stations that use google map . in particular , city_info stores the town in which each point ( station ) is located , along with additional information .
fields include id ( main key ) , title , en_title , lat , log .
table points_categories stores the station s category ( fields : id is the primary key , title and en_title ) . on the table
points points are set up , along with additional information .
these fields are id ( primary key ) , category , city , en_city , address , title , description , lat , log , thumb , image .
finally 12 tables ( mamyntaio , mflorina , mgrevena , mkastoria , mkkomi , mkozani , mnpedio , mpetrana , mpontokomi , mptl , mservia , msiatista ) which store weather forecast information are used .
every table represents a certain location in western macedonia .
fields date , hour have been set up as primary key denoting solely each record .
the rest ( wdir , temp , rhum , tempscr , rhumscr , tsr , netr , sens , evap , wstar , zmix , ustar , lstar , rain and snow ) are meteorological parameters .
tables about pollution forecasts do not exist , all measurements are stored in the hard disc of the server as image files .
a sample table of weather forecast has the following form : mkozani ( date , hour , wdir , temp , rhum , tempscr , rhumscr , tsr , netr , sens , evap , wstar , zmix , ustar , lstar , rain and snow . )
the proposed a.q.m.e.i.s .
application is part of a system - air quality monitoring network , which was developed in the laboratory of atmospheric pollution and environmental physics of technological education institute of western macedonia , to monitor the air quality in western macedonia area , with industrial focus on the region of prolemais - kozani basin .
this system was co - financed by the teiwm , regional operational programm 2000 - 2006 western macedonia and recently by the municipality of kozani .
the architecture of this system is constituted by five terminal stations , which collect environmental information , the central station and a web server .
different technologies ( adsl , gprs , ethernet ) are used to transfer the data to the central station .
the data are sent every half an hour to the main station which collects the complete set of data and transfers them to the web server every sixty minutes , where under the application proposed in this paper provides meteorological , environmental , weather and air pollution forecast data in west macedonia area .
further details on the design of the above mentioned air quality monitoring network can be obtained from @xcite,@xcite , @xcite .
an operational monitoring , as well as high resolution local - scale meteorological and air quality forecasting information system(a.q.m.e.i.s . ) for western macedonia , hellas , has been developed and is operated by the laboratory of atmospheric pollution and environmental physics / tei western macedonia . in this paper the novelty of information system is presented , in a dynamic , easily accessible and user - friendly manner . the application is developed using state of the art web technologies ( ajax , google maps etc ) and under the philosophy of the open source software that gives the ability to users / authors to update / enrich the code so that their augmentative needs are met .
design of a webbased information system for ambient air quality data in west macedonia , greece , a.g .
triantafyllou , v. evagelopoulos , e.s .
kiros , c. diamantopoulos , 7th panhellenic ( international ) conference of meteorology , climatology and atmospheric physics , nicosia 28 - 30 september , 2004 .
council directive 90/313/eec of 7 june 1990 on the freedom of access to information on the environment .
council directive 92/72/eec of 21 september 1992 on air pollution by ozone .
council directive 1999/30/ec of 22 april 1999 relating to limit values for sulphur dioxide , nitrogen dioxide and oxides of nitrogen , particulate matter and lead in ambient air .
council directive 96/62/ec of 27 september 1996 on ambient air quality assessment and management .
council directive 90/313/ec in expanding the level of access in information karatzas k. and j. kukkonen , 2009 , cost action es0602 workshop proceedings `` quality of life information services towards a sustainable society for the atmospheric environment '' ed .
k. karatzas and j. kukkonen .
schimak g. , 2003 : environmental data management and monitoring system uwedat .
environmental modelling and software 18 , 573 - 580 .
xuan zhu , allan p. dale , 2001 : javaahp : a webbased decision analysis tool for natural resource and environmental management , environmental modelling and software 16 , 251 - 262 . atmospheric pollution - atmospheric boundary layer , advances in measurements technologies , a.g .
triantafyllou , ed .
tei of western macedonia , kozani 2004 , pp.256 .
triantafyllou , v. evagelopoulos , s. zoras , design of a webbased information system for ambient air quality data , journal of environmental management , 80(3 ) , 230 - 236 ( 2006 ) .
triantafyllou , j. skordas , ch.diamantopoulos , e. topalis , `` the new dynamic air quality information systemand its application in west macedonia , greece '' , p.4 , 4th macedonian enviromental conference , thessaloniki , greece , march 2011 .
ioannis skordas , george f. fragulis , athanasios g. triantafyllou , `` eairquality : a dynamic web application for evaluating the air quality index for the city of kozani , hellas '' , pci 2011 15th , panhellenic conference on informatics , 30 september - 02 october 2011 , kastoria , greece .
http://www.airlab.edu.gr/ beginning web programming with html , xhtml , and css , jon duckett , wiley publishing , 2008 .
core php programming 3rd edition , leon atkinson with zeev suraski , pearson education , inc , publishing as prentice hall professional technical reference , 2004 mysql : the complete reference , vikram vaswani , mcgraw - hill brandon a. nordin , 2004 .
apache : the definitive quide 3rd edition , ben laurie and peter laurie , oreilly , 2003 .
triantafyllou a.g .
, krestou a. , hurley p. and thatcher m. , `` an operational high resolution localscale meteorological and air quality forecasting system for western macedonia , greece : some first results '' , proceedings of the 12th international conference on environmental science and technology ( cest2011),8 10 september 2011 , pp.a1904:1911 | an operational monitoring , as well as high resolution local - scale meteorological and air quality forecasting information system for western macedonia , hellas , has been developed and is operated by the laboratory of atmospheric pollution and environmental physics / tei western macedonia since 2002 , continuously improved . in this paper the novelty of information system is presented , in a dynamic , easily accessible and user - friendly manner .
it consists of a structured system that users have access to and they can manipulate thoroughly , as well as of a system for accessing and managing results of measurements in a direct and dynamic way .
it provides updates about the weather and pollution forecast for the next few days ( based on current day information ) in western macedonia .
these forecasts are displayed through dynamic - interactive web charts and the visual illustration of the atmospheric pollution of the region in a map using images and animation images . |
high - velocity outflows from quasars appear to be a natural part of the accretion process .
they are detected most conspicuously via broad absorption lines ( bals ) in quasar spectra , with velocity widths @xmath102000 km s@xmath1 ( by definition , * ? ? ?
* ) and flow speeds from a few thousand to roughly 10,000 - 30,000 km s@xmath1 ( see also * ? ? ?
* ; * ? ? ?
however , quasar outflows can also produce narrow absorption lines ( nals ) , with velocity widths less than a few hundred km s@xmath1 , as well as mini - bals , which are loosely defined as intermediate between nals and bals ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these narrower outflow lines are more common than bals @xcite .
overall it is estimated that @xmath1050% of optically - selected quasars have at least one type of outflow absorption line in their rest - frame uv spectra @xcite .
the flows themselves are probably present in all quasars if , as expected , the outflowing gas covers just a fraction of the sky as seen from the central emission source ( e.g. , * ? ? ?
quasar outflows have gained attention recently as a mechanism that can physically couple quasars to the evolution of their host galaxies .
in particular , some models of galaxy evolution invoke the kinetic energy feedback " from an accreting super - massive black hole ( smbh ) to regulate both the star formation in the host galaxies and the infall of matter toward the central smbh ( * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and refs .
therein ) . this type of coupling via feedback could provide a natural explanation for the observed mass correlation between smbhs and their host galaxy spheroids @xcite .
the kinetic energy luminosity , @xmath11 , needed for significant feedback from a quasar outflow is believed to be just a few percent of the quasar bolometric ( photon ) luminosity , e.g. , @xmath12 @xcite .
unfortunately , many aspects of the energetics and physical properties of quasar outflows remain poorly understood .
computational models aimed mainly at bals attribute the flows to gas lifted off of the accretion disk and driven outward to high speeds by radiation pressure or magneto - centrifugal forces @xcite .
radiative forces are expected to dominate in quasars , where the bolometric luminosities are a substantial fraction of the eddington luminosity @xcite .
one essential requirement for radiative driving is that the outflow is not too highly ionized , so it can maintain significant line and bound - free continuum opacities across the main acceleration region .
this is not trivial for flows launched from the inner accretion disk , near the quasar s intense source of ionizing radiation .
@xcite proposed that the outflow ionization is moderated by a high column density of roughly stationary material that serves as a radiative shield at the base of the flow ( also * ? ? ?
the shielding medium is itself too ionized and too transparent to be driven radiatively to high speeds , but it might have enough continuous opacity in the far - uv and x - rays to lower the ionization and thereby facilitate the acceleration of the gas directly behind it ( i.e. , farther from the emission source ) .
this scenario is supported by observations showing that bal quasars are heavily obscured in x - rays , consistent with strong radiative shielding @xcite . however ,
other observations indicate that quasars with nal or mini - bal outflows have substantially less x - ray absorption @xcite .
this disparity in the x - ray results , between bals on the one hand and nals and mini - bals on the other , presents an important challenge to the paradigm of radiative acceleration behind an x - ray / far - uv shield ( see also 4.5 below ) . a central problem is our poor understanding of the physical relationships between bal , nal and mini - bal outflows .
one possibility is that the different line types are simply different manifestations of a single outflow phenomenon viewed at different angles . for example
, it is often supposed that bals form in the main body of the outflow near the accretion disk plane , while nals and mini - bals form along sightlines that skim the edges of the bal flow at higher latitudes above the disk @xcite .
there could also be evolutionary effects .
for example , nals and mini - bals might represent the tentative beginning or end stages of a more powerful bal outflow phase @xcite .
there is some evidence for outflow evolution in that a particular variety of low - ionization bals ( the so - called felobals ) is found preferentially in dusty young host galaxies with high star formation rates @xcite .
variability studies have shown further that bals and mini - bals in quasar spectra can appear , disappear or swap identities ( where a bal becomes a mini - bal or vice versa ) on time scales of months to years in the quasar rest frame @xcite .
the true relationship between outflow nals , bals and mini - bals probably involves an amalgam of orientation and time - dependent effects . to build a more complete picture of quasar outflows ,
we need better observational constraints on the physical properties of each outflow type .
nal outflows present a unique challenge because they are difficult to identify .
most narrow absorption lines in quasar spectra form in cosmologically intervening gas that has no relationship to the background quasar .
statistical studies that examine the velocity distributions of the narrow metal lines @xcite , or correlations between the incidence of these lines and the properties of the background quasars @xcite , indicate that significant fractions of the narrow metal - line systems do arise in quasar outflows .
however , to identify individual nal outflow systems we need more information .
the most commonly used indicators of an outflow origin are line variability , resolved absorption profiles that are significantly broad and smooth compared to thermal line widths , or line strength ratios in multiplets that reveal partial line - of - sight covering of the background light source @xcite . here
we discuss a complex of five outflow nal systems at velocities from @xmath09710 to @xmath014,050 km s@xmath1 in the redshift @xmath82.3 quasar j212329.46 - 005052.9 ( hereafter j2123 - 0050 ) .
to our knowledge , these are the highest velocity nals whose outflow nature is confirmed by all three indicators of variability , partial covering and smooth super - thermal line profiles .
we selected this quasar from the sloan digital sky survey ( sdss ) as part of a larger study of intervening , metal - strong damped ly@xmath13 ( dla ) and sub - dla absorption line systems @xcite .
follow - up observations @xcite revealed the variability and outflow origin of the nal systems discussed here . in the sections below
we describe the observations ( 2 ) , present measurements and analysis of the outflow lines ( 3 ) , and discuss the interpretation and broader implications of our results ( 4 ) .
we provide a detailed general summary in 5 . throughout this paper ,
we adopt a cosmology with @xmath14 km s@xmath1 mpc , @xmath15 and @xmath16 .
table 1 summarizes the observations used in this study .
columns 1 and 2 give the fractional years and calendar dates of the observations .
columns 3 - 5 list the telescope , spectral resolution and wavelength coverage , @xmath17 , respectively .
we obtained the 2002.68 spectrum from the sdss archives fully reduced and calibrated .
the fluxes provided by the sdss are expected to be accurate to within a few percent @xcite .
unfortunately , these are the only reliable absolute fluxes available for our spectra .
we observed j2123 - 0050 at the mmt observatory using the mmt spectrograph in 2006.71 and 2009.40 . the resolutions and wavelength coverages listed in table 1
were obtained using a 1.5@xmath18 wide entrance slit together with the 800 groove / mm grating in the blue channel .
we extracted and calibrated these spectra with standard techniques using the low redux pipelinexavier / lowredux / index.html ] software package .
we performed relative flux calibrations using spectrophotometric standards observed on the same night .
the emission - line redshift of j2123 - 0050 reported by the sdss , @xmath19 , was modified recently to @xmath20 by @xcite .
these redshifts match the peak position of the broad @xmath2@xmath21548,1551 emission line quite well .
however , the lower ionization lines @xmath21303 and @xmath21336 appear at a slightly higher redshift that is probably closer to the true quasar systemic ( within @xmath8100 km s@xmath1 ; * ? ? ? * ; * ? ? ?
we therefore estimate the redshift from the and lines as follows .
first we average together the sdss and mmt 2006.71 spectra to maximize the signal - to - noise ratio .
then we manually interpolate across the tops of any narrow absorption lines to eliminate these features ( e.g. , in the emission line profile ) , and calculate the and emission line centroids .
this yields an average redshift for the two lines of @xmath21 ( using laboratory wavelengths from * ? ? ?
measurements based on the line peaks or the centroids of only the upper halves of the profiles yield essentially the same result . also , redshifts determined separately for the two lines agree within the measurement uncertainty quoted above .
we will adopt the redshift @xmath21 for the quasar systemic throughout the remainder of this paper .
.observation summary [ cols="<,<,^,^,^ " , ] next we transfer the fits to other lines in the variable systems by keeping the kinematic parameters @xmath22 , @xmath23 and @xmath24 fixed while scaling only @xmath25 and @xmath26 , as needed . for system
b , we adopt @xmath27 km s@xmath1 and @xmath28 km s@xmath1 based on our efforts to add a system b contribution to the blend . this simple shift - and - scale approach is essential for dealing with non - detections and blends in the forest .
it is also justified by the good results shown for and in figures 4 - 6 . however , as noted in 3.2 above , the absorption appears to span a wider range of velocities than .
our fits ignore this additional absorption because i ) it is poorly measured due to blending problems , and ii ) we want to limit our comparisons between and to the same kinematic gas components anyway , for the ionization and abundances analysis in 3.6 and 3.7 below . table 4 lists the fit parameters derived for the other lines .
the doublets measured in 2006.64 and 2008.67 provide useful constraints on both @xmath26 and @xmath29 .
notice that the covering fractions are larger than ( c.f . , table 3 ) . also note that several of the doublets are saturated , with intensity ratios close to unity ( see figs . 4 and 5 ) .
for these systems , table 4 lists lower limits on the column densities that correspond to @xmath30 in the stronger @xmath21032 line .
the values of @xmath31 listed for system c and system e in 2008.67 ( only ) come from fits indicating @xmath32 .
however , we could conservatively consider all of the column densities to be lower limits because of the blending problems in the forest .
we estimate 1@xmath33 uncertainties in the results by repeating the fits with different continuum placements while keeping @xmath34 and the doppler parameters fixed , as described above .
we find that the errors in @xmath26 are @xmath3515% while the errors in @xmath31 are roughly 30% .
we conclude that the differences in the derived covering fractions between and , and the smaller changes in the covering fractions compared to the changes in ( c.f . , tables 3 and 4 ) , both have a large significance .
@ccccc@ line & system & year & @xmath26 & @xmath31 + & & & & ( 10@xmath36 @xmath37 ) + & a & 2006.64 & 0.95 & @xmath1058.4 + & a & 2008.67 & 0.82 & @xmath1053.5 + & b & 2006.64 & 0.38 & @xmath1058.4 + & b & 2008.67 & 0.30 & @xmath1058.4 + & c & 2006.64 & 0.87 & 51.1 + & c & 2008.67 & 0.82 & 55.4 + & d & 2006.64 & 0.50 & @xmath10136 + & d & 2008.67 & 0.50 & @xmath10102 + & e & 2006.64 & 0.54 & @xmath10136 + & e &
2008.67 & 0.63 & 92.0 + + & a & 2006.64 & & @xmath350.30 + & b & 2006.64 & & @xmath350.30 + & c & 2006.64 & & @xmath350.30 + & d & 2006.64 & & @xmath350.60 + & e & 2006.64 & & @xmath350.60 + + & a & 2006.64 & & + & b & 2006.64 & & @xmath358.5 + & c & 2006.64 & & @xmath3518 + & d & 2006.64 & & + & e & 2006.64 & & @xmath3572 + + & a & 2006.64 & & @xmath355.2 + & b & 2006.64 & & @xmath350.6 + & c & 2006.64 & & 4.8 + & c & 2008.67 & & 7.7 + & d & 2006.64 & &
@xmath356.4 + & e & 2006.64 & & @xmath3523 + + & a & 2006.64 & & @xmath355.0 + & b & 2006.64 & & @xmath353.0 + & c & 2006.64 & & @xmath353.0 + & d & 2006.64 & & @xmath3515 + & e & 2006.64 & & @xmath3515 + the remaining lines listed in table 4 are either non - detections ( and ) or severe blends ( and ) where we can not use line intensity ratios to determine both @xmath25 and @xmath26 .
we therefore fix @xmath26 along with the kinematic parameters and scale only @xmath25 to derive constraints on @xmath31 .
figure 6 shows examples of these fits applied to in systems a , b and c. we adopt values of @xmath26 from ( table 3 ) in all cases except for in system b , which has no fit result , and , which has measured lines / blends too deep to be consistent with @xmath26 in . for these latter cases
we adopt the covering fractions from .
two of the entries for in table 4 are blank because the blending is too severe to yield meaningful results .
the remaining three systems ( b , c and e ) are listed as upper limits in table 4 because of blending uncertainties .
we note , however , that there do appear to be real lines present with redshifts and profiles similar to .
the upper limits on @xmath31 listed for in table 4 use the best available constraints from either or .
the specific values of the column density given for system c come from the fits shown in figure 6 without additional constraints from .
the errors in these results might be as large as a factor of @xmath82 due mainly to the uncertainties in @xmath26 . as a further check of our fitting analysis , we derive velocity - dependent values of @xmath38 and @xmath39 for the two best - measured cases : the lines in systems d and e observed in 2006.64 .
specifically , we calculate the average line intensities in velocity bins 25 km s@xmath1 wide and then solve equation 1 for @xmath38 and @xmath39 in each bin .
figure 7 compares the derived values of @xmath40 to the observed line profiles .
the vertical bars running through the filled circles in figure 7 represent 1@xmath33 uncertainties in @xmath39 caused by photon statistics @xcite .
however , in the blue wings of both systems d and e there are several velocity bins where the doublet ratios indicate @xmath41 and the standard analysis yields @xmath42 .
the actual values of @xmath38 and @xmath39 are poorly constrained in these cases because the right side of equation 1 reduces to @xmath43 in the @xmath41 limit and , therefore , the line intensities constrain only the product @xmath44 instead of @xmath39 and @xmath38 separately . for example , in the velocity bin centered at @xmath45 km s@xmath1 in system d , the data are consistent with solutions ranging from @xmath46 and @xmath47 to @xmath48 and @xmath49 .
figure 7 shows both the maximum and minimum ( @xmath50 ) results , connected by a vertical bar , for each of the optically thin data points .
the data at these velocities are compatible with any value of @xmath40 within the ranges shown .
( filled circles ) , on a velocity scale relative to @xmath51 .
the solid and dashed curves show the stronger ( @xmath21548 ) and weaker ( @xmath21551 ) doublet members , respectively , after smoothing the spectra 3 times with a binomial function .
the vertical bars on the filled circles are 1@xmath33 uncertainties . for the velocity bins with @xmath41 ( in the blue line wings ) ,
two filled circles at each velocity represent the full range of @xmath40 values consistent with the data . ]
the values of @xmath39 and @xmath38 derived from the point - by - point analysis near line center agree well with @xmath26 and @xmath25 obtained above from the gaussian fits ( table 3 ) .
the point - by - point results in figure 7 show further that @xmath39 changes significantly across the line profiles . in particular
, the shape of the intensity profile across the core of system e appears to be governed largely by the behavior of @xmath39 with @xmath52 . in system d , there appears to be an abrupt change in @xmath39 around @xmath013,200 km s@xmath1 , between a line core that has @xmath53 and @xmath54 and a blue wing with @xmath41 and @xmath55 . integrating @xmath38 over all of the velocity bins shown in figure 7 ( as per equation 2 but excluding the lowest @xmath56 point in the red wing of both systems because they formally yield @xmath57 due to noise ) indicates @xmath58 @xmath37 for system d and @xmath59 @xmath37 for system e. these column densities are in excellent agreement with the gaussian fit results in table 3 .
we conclude with a comment about the assumption of spatial uniformity in the absorbing regions .
some studies of partial covering in quasar absorption lines indicate that the the actual absorbing structures are inhomogeneous , with a range of @xmath38 values at each @xmath56 across the projected area of the emission source ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and refs .
therein ) .
the evidence for this is significantly different covering fractions in different lines / ions . while these data can provide important constraints on the nature of the absorber inhomogeneities ,
there is not enough information to determine uniquely the 2-dimensional spatial distributions of @xmath38 in each line .
the situation is further complicated by the fact that the background light sources , e.g. , the accretion disk and/or the belr , have their own spatial non - uniformities that can include a wavelength - dependent size .
a good strategy , therefore , is to consider the observational consequences of different plausible @xmath38 spatial distributions @xcite . in one study , extensive simulations using different functional forms for the @xmath38 spatial distribution show that equation 1 ( which assumes a top hat @xmath38 distribution ) yields good approximations to the spatially - averaged values of @xmath38 and @xmath31 in more general inhomogeneous absorbing media @xcite .
therefore , the results derived from equation 1 are appropriate for making spatially - averaged estimates of the ionization and metal abundances ( 3.6 below ) .
they are also useful for characterizing the covering fractions even for complex absorbers .
in particular , the values of @xmath39 derived from equation 1 correspond roughly to the amount of coverage by absorbing material with @xmath60 in an inhomogeneous medium @xcite .
the partial covering discussed in 3.3 implies that the absorbing structures are not much larger than the background light source .
( the absorbers could be smaller than the emission source , but partial covering by much larger clouds would require unphysically sharp edges and extremely unlikely spatial alignments . )
figure 2 shows that the variable systems lie in the far blue wing of the bel , where the flux is dominated by continuum emission . in principle
, the line absorption could occur in front of both the uv continuum source and the belr , with potentially different covering fractions for each region @xcite .
however , the depths of the lines below the quasar continuum ( figure 2 ) imply that the partial covering results pertain mostly or exclusively to the continuum source .
this conclusion is confirmed by the lines , which have even deeper absorption troughs and no significant underlying belr flux . the extent to which the absorbers cover the belr are thus undetermined by our data .
this ambiguity introduces a small additional uncertainty in @xmath39 derived for the lines because the belr contributes roughly 10% of the total flux at those wavelengths ( figure 2 ) . our analysis in 3.3 assumed implicitly that @xmath39 is the same for all emission sources beneath the absorption lines .
however , if the belr is not covered at all , the continuum covering fractions would be roughly 10% larger than the values listed for in table 3 .
conversely , if the absorber completely covers the belr , the continuum covering fractions would be @xmath820% smaller than the results listed in table 3 ( see equations in * ? ? ?
this latter situation ( complete coverage of the belr with partial covering of the continuum source ) seems highly unlikely given that the belr is of order 20 times larger than the continuum source at 1550 ( 2.1 ) .
hereafter , we will assume that all of the derived covering fractions apply only to the continuum source .
comparing the covering fractions in and ( tables 3 and 4 ) to the theoretical diameters of the continuum source at 1550 and 1034 , respectively ( 2.1 ) , indicates that the and absorbing regions both have characteristic sizes of order @xmath80.01 - 0.02 pc ( as measured in 2006.64 ) .
it is important to keep in mind , however , that this characteristic size is actually just an upper limit on the true sizes of absorbing structures in the flow .
in particular , partial covering might be caused by patchy distributions of many smaller sub - structures ( see 4.4 below , also * ? ? ?
the five distinct outflow systems span a range of velocities from @xmath09710 to @xmath014,050 km s@xmath1 and line widths from fwhm @xmath3 62 to 164 km s@xmath1 ( table 2 ) .
the lines , representing higher ionization material , appear somewhat broader than in terms of both their fwhms and broad weak wings ( see figs . 4 and 5 ) .
the outflow systems have a natural division into two groups , a+b+c and d+e , based on similarities in the line strengths , profiles and velocity shifts ( see also fig .
one remarkable characteristic of these systems overall is their small velocity dispersions compared to the high flow speeds , e.g. , fwhm/@xmath61 .
this in marked contrast to bal outflows , which typically have fwhm/@xmath62 @xcite .
the kinematic stability of the narrowest systems a , b and c between 2006.64 and 2008.67 places an upper limit on the radial acceleration of @xmath53 km s@xmath1 yr@xmath1 in the quasar frame .
we obtain this estimate from the measurements in table 3 and by careful comparisons of simultaneous fits to all three narrow systems ( in and ) in both epochs .
( the centroids of the broader systems d and e did change slightly between observations , but this appears to be the result of profile changes rather than a true velocity shift ; figure 5 . ) for comparison , the gravitational acceleration due to @xmath63 at the radius of the belr ( 2.1 ) is roughly 200 km s@xmath1 yr@xmath1 .
therefore , the outward driving force is either in a delicate balance with gravity or , more likely , the absorbing gas is located in a low - gravity environment farther from the black hole than the belr . a radial distance of @xmath64 pc is needed for the local gravity to be less than the observed acceleration limit . at this distance
, the observed outflow speeds vastly exceed the local gravitational escape speed .
thus it appears that the absorbing gas is gravitationally unbound and coasting freely at speeds near the outflow terminal velocity .
there is an apparent occurrence of line - locking " between the narrow systems a , b and c. in particular , the velocity differences between systems a - b and b - c are a close match to the doublet separation of @xmath65 km s@xmath1 .
( although the system b lines are entirely blended in , their velocity is confirmed by our fits to the lines shown in figure 4 . )
line - locking is usually interpreted as evidence for radiative acceleration @xcite .
cases of multiple line - locks between systems with narrow profiles , as in j2123 - 0050 , provide the best evidence for physically locked systems because they are difficult to explain by chance alignments of lines in the spectrum .
however , it is not known if radiative forces can actually lock doublets like together in real quasar outflows .
the most plausible scenario was outlined by @xcite : consider two blobs of gas moving along the same radial path in a radiatively accelerated outflow .
the faster blob is farther from the emission source and experiencing a _
greater _ outward acceleration .
when the velocity difference between the blobs matches the doublet separation , @xmath66 km s@xmath1 , the @xmath21551 line in the outer blob falls in the shadow cast by @xmath21548 in the slower inner blob .
this leads to diminished acceleration in the outer blob and a velocity lock between the two blobs at the doublet separation .
line - locking might even contribute to the formation of blobs because the localized dip in the radiative force downstream from the inner blob creates a natural collecting area for outflow gas @xcite .
this scenario does not require that the lines dominate the overall acceleration , only that the acceleration transferred to the outer blob via the @xmath21551 line is at least as large as the acceleration _ difference _ between the two blobs in the absence of shadowing ( i.e. , outside of the @xmath66 km s@xmath1 line overlap situation ) .
the appearance of a double line - lock in , as in j2123 - 0050 , might also be facilitated by the absorption doublet whose separation , @xmath67
km s@xmath1 , is close to the velocity difference between systems a and c. in any case , the feasibility of this line - lock interpretation has not been demonstrated in any quasar outflow . here
we note simply that if radiative forces did indeed lock the doublets in j2123 - 0050 , then the absorbing clouds must be moving almost directly toward us .
this is because the velocity difference between line - locked blobs should match the doublet separation along the flow direction .
if the flow is not aimed directly at us , we should observe velocity differences between the systems that are smaller than the doublet separation .
our measurements of the line centroids ( table 2 ) and our fits to the profiles , including in system b ( table 3 and figure 4 ) , indicate that the velocity differences between systems a , b and c are within @xmath820 km s@xmath1 of @xmath65 km s@xmath1 .
this implies that the absorber trajectories are within @xmath816@xmath9 of the purely radial ( line - of - sight ) direction .
the strong detections of compared to absorption , and the non - detections of lower ionization lines like and ( tables 3 and 4 ) , indicate that the degree of ionization is high and approximately the same in all of the variable systems .
the appendix below describes numerical simulations using the code cloudy @xcite of the ionization and physical conditions in clouds that are in photoionization equilibrium with a quasar continuum source . comparing those results ( figure a1 ) to the most stringent limit on the observed @xmath68 column density ratio ( system c in 2006.64 )
indicates that the dimensionless ionization parameter is at least @xmath69 .
the measured ratios of @xmath70 provide a very similar constraint on @xmath71 if we make the additional assumption that the o / c abundance is approximately solar .
this lower limit on the ionization combined with the average measured column density in system c , @xmath72 @xmath37 ( table 4 ) , implies that the minimum total hydrogen column density in that system is @xmath73 @xmath37 ( see figure a1 ) . in
4.3 below , we argue that the line changes were caused by changes in the degree of ionization and the most likely value of the ionization parameter is @xmath74 , near the peak in the ion fraction . using that result , our best estimate of the actual total column density in system c is of order @xmath75 @xmath37 . given the similarities between the systems a - e , we crudely estimate that the total column density in all five outflow absorbers is @xmath76 @xmath37 . the more conservative requirement for @xmath77 in a photoionized gas at @xmath64 pc ( 3.5 ) from the quasar places an upper limit on the volume density of @xmath78 @xmath79 ( see equation a1 )
. a lower limit on the density can be derived by noting that , in photoionization equilibrium , changes in the ionization state require at least a recombination time given by @xmath80 , where @xmath81 is the electron density and @xmath82 is the recombination rate coefficient . given @xmath83
@xmath84 s@xmath1 for either @xmath85 or @xmath85 at a nominal gas temperature of @xmath86 k ( see * ? ? ?
* ; * ? ? ?
* ) , the observed variability in @xmath40.63 yr indicates a minimum density of @xmath87 @xmath79 . this density combined with the ionization constraint @xmath77 yields a maximum distance of @xmath88 kpc ( eqn .
the density and distance constraints based on the recombination time ( @xmath89 @xmath79 and @xmath88 kpc ) might not apply if the ionization changes occurred out of equilibrium .
in particular , ionization _ increases _ caused by a rise in the ionizing flux could occur much faster than the recombination time .
the most likely explanation for the line changes between 2006.64 and 2008.67 is , indeed , an ionization increase ( 4.3 below ) .
however , there is weak evidence for both weakening and strengthening of the lines in j2123 - 0050 ( fig .
3 ) , which is typical variable outflow lines in quasars @xcite . if this up and down behavior of the line strengths represents changes in the ionization ( see also * ? ? ?
* ) , then recombination must be involved
. continued high - resolution monitoring is needed to confirm that conclusion specifically for j2123 - 0050 .
however , even if the line variations in this object occurred because of clouds crossing our view of the continuum source , basic assumptions about the crossing speeds would require a radial distance much smaller than the 1.1 kpc limit derived from the recombination time ( 4.3 ) .
therefore , we adopt @xmath88 kpc as a firm upper limit .
we estimate the c / h metal abundance from the average measured ratio of @xmath90 in system c ( tables 3 and 4 ) and an ionization correction based on @xmath74 ( see eqn . a2 and fig .
this indicates [ c / h ] @xmath91 , where the square brackets have their usual meaning of log abundance relative to solar and the 1@xmath33 uncertainty includes an estimate of the errors in both the column densities and the ionization correction .
the uncertainties related to the ionization are small because the most likely ionization parameter , @xmath74 , is near the peak in the ion fraction curve where the correction factor , @xmath92 , is not sensitive to the specific value of @xmath71 across a fairly wide range .
a careful inspection of the numerical results depicted in figure a1 shows that @xmath92 stays constant within @xmath930.1 dex across a range in @xmath71 from @xmath01.4 to 0.0 .
moreover , this range in @xmath71 encompasses the minimum value of @xmath92 at @xmath94 ( table a1 ) .
if we apply the minimum correction factor @xmath95 ( table a1 ) to the system c column densities , we derive a lower limit on the metal abundance , [ c / h ] @xmath96 , which is close to our best estimate above but completely independent of the ionization uncertainties ( see also * ? ? ?
section 3 presented measurements and analysis of five distinct outflow nal systems in j2123 - 0050 having line widths of fwhm @xmath8 62 to 164 km s@xmath1 and velocity shifts from @xmath97 to @xmath014,050 km s@xmath1 in the quasar rest frame .
these systems appear to be physically related based on their roughly similar line strengths , kinematics , ionizations , line - of - sight covering factors , and coordinated variabilities .
they provide a wealth of information about a particular nal outflow that is valuable for comparison to other work on bal and mini - bal outflows . here
we discuss some additional results and implications .
please see 5 below for an overall summary .
the formation of the variable nals in a quasar outflow is confirmed by all three properties of line variability , partial covering and resolved profiles that are smooth and broad compared to thermal line widths . moreover ,
if we accept that the lines form somewhere in the quasar environment based on these properties , then the only plausible location is a quasar - driven outflow because the velocity shifts are much too large to be explained by other near - quasar environments such as a galactic starburst - driven wind or a nearby galaxy in the same cluster as the quasar @xcite . by matching the variability to a recombination time , we place an upper limit on the distance between the absorber and the quasar of @xmath88 kpc ( 3.6 ) . at the opposite extreme , constraints on the acceleration suggest that the nal gas is coasting freely at a distance @xmath98 pc ( 3.5 ) .
thus the full range of plausible distances is @xmath7 pc .
however , if the absorbing structures are small blobs or filaments created in the inner flow , their most likely location is near the @xmath85 pc minimum radius because such structures will travel just a few pc before dissipating ( if there is no external confinement . 4.4 below ) . finally
, our estimate of the metallicity , [ c / h ] @xmath91 with a lower limit of [ c / h ] @xmath96 , is also consistent with the formation of these nals in a quasar outflow .
super - solar metallicities are extremely rare in intervening absorption line systems , having been observed so far only in a few lyman - limit and super - lyman - limit systems @xcite , but they are typical of near - quasar environments based on estimates from the broad emission lines @xcite and from other well - studied cases of nal outflows ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* simon , hamann & pettini 2010 , in prep . ) .
it is interesting to note that the outflow nals in j2123 - 0050 are unresolved and indistinguishable from cosmologically intervening lines in medium resolution spectra like the sdss .
they are much too narrow to be identified with an outflow using one of the absorption line indices designed for this purpose @xcite .
their velocity shifts are also well above the nominal @xmath99 km s@xmath1 cutoff used to define `` associated '' absorption lines ( aals ) , which are likely to have a physical relationship to the quasar based on their statistical excess at @xmath100 compared to @xmath101 @xcite . without high resolution spectra or multi - epoch observations to test for variability , the outflow nals in j2123 - 0050 would be mistakenly attributed to cosmologically intervening gas .
the outflow systems are , in fact , surrounded in the j2123 - 0050 spectrum by other narrow lines that did not vary , do not have partial covering and almost certainly do form in unrelated intervening material ( figure 1 ) .
one of these intervening systems is at a smaller velocity shift than the outflow lines ( at @xmath102 km s@xmath1 ) , while another is blended directly with the outflow system e ( see also table 2 and figure 5 ) .
this mixture of lines in the same spectrum clearly demonstrates that velocity shift alone is a poor indicator of the outflow versus intervening origin of narrow absorption lines ( see also * ? ? ?
the secure outflow origin of the nals in j2123 - 0050 lends anecdotal support to recent claims that a significant fraction of narrow systems , even at large velocity shifts , form in quasar outflows .
for example , @xcite argued that @xmath836% of high - velocity nals originate in outflows based on a correlation between the numbers of these systems detected and the radio properties of the background quasars .
@xcite estimated that 10 - 17% of narrow high - velocity systems belong to quasar outflows based on the incidence of partial covering measured in high - resolution spectra .
similar results were obtained more recently by simon et al .
( 2010 , in prep . ) .
@xcite showed that the statistical excess of nals near the quasar redshift extends out to at least @xmath812,000 km s@xmath1 , e.g. , well beyond the nominal cutoff of 5000 km s@xmath1 for `` associated '' absorption lines .
they argue that @xmath643% of nals stronger than rew = 0.3 in the velocity range @xmath103 km s@xmath1 originate in quasar outflows ( also * ? ? ?
* ) . in their analysis of mostly weaker systems , simon et al .
( 2010 , in prep . )
find that roughly 20 - 25% of nals in this velocity range exhibit partial covering .
the variable lines in j2123 - 0050 are unique in that they are ( so far ) the highest velocity nals known to originate in a quasar outflow based on all three indicators of line variability , partial covering and super - thermal line widths .
their closest analogues in the literature are a complex of nals at somewhat lower speeds , @xmath104 km s@xmath1 , in the redshift 2.5 quasar hs 1603 + 3820 @xcite , and an isolated broader ( fwhm @xmath105 km s@xmath1 ) outflow system at @xmath106 km s@xmath1 in the @xmath107 quasar q2343 + 125 @xcite .
the key properties of the nal variability in j2123 - 0050 are 1 ) the line variations were well coordinated between the five systems , each with larger changes in than , 2 ) changes in the nal strengths were accompanied by roughly commensurate changes in the line - of - sight covering fractions , and 3 ) the variability time is @xmath40.63 yr in the quasar rest frame .
changes in the covering fractions would seem to have a natural explanation in clouds moving across our lines of sight to the background emission source .
this explanation is , in fact , favored by some recent studies of bal variability @xcite .
however , in j2123 - 0050 , line changes induced by crossing clouds appear unlikely because the coordinated variations in five distinct nal systems would require highly coordinated movements between five distinct absorbing structures in the outflow ( also * ? ? ?
moreover , crossing clouds would need to traverse a significant fraction of the emission source diameter in @xmath40.63 yr to cause the observed line variations . in
3.4 we argued that the relevant emission source is the accretion disk with diameter @xmath108 pc at 1550 . crossing @xmath10915% of that length in @xmath40.63 yr , to explain the observed changes of order 0.15 in the covering fractions , would require transverse speeds @xmath110 km s@xmath1 .
this seems unrealistic given that the radial flow speeds are not much larger , e.g. , @xmath97 to 14,050 km s@xmath1 .
the maximum transverse speeds we might expect in a quasar outflow launched from a rotating accretion disk are of order the virial speed at the absorber location .
a virial speed consistent with @xmath111 km s@xmath1 in j2123 - 0050 would require that the absorber resides at a radial distance @xmath52.4 pc from the smbh .
this would place the absorbers inside of the minimum radius of 5 pc estimated above from the lack of radial acceleration ( 3.5 ) .
it would also contradict the evidence from line - locking that the flow we observe is traveling primarily in a radial direction ( 3.5 ) .
a more likely explanation for the coordinated line variations is global changes in the ionization caused by fluctuations in the quasar s continuum flux .
this situation could produce the observed changes in the covering fractions ( 3.3 ) if the absorbers are spatially inhomogeneous , i.e. , with a distribution of column densities across the emission source , such that the projected areas with @xmath112 in a given line change with the ionization state of the gas @xcite .
the observed decrease in the line strengths between 2006.64 and 2008.67 , accompanied by smaller or negligible changes in , are consistent with an overall increase in the ionization parameter if the starting point was near the peak in the ion fraction .
in particular , an ionization parameter initially near @xmath74 in 2006.64 and then increasing by a few tenths of a dex by 2008.67 would cause a decline in the line strengths and column densities consistent with the observations ( table 3 ) , while the remains nearly constant ( see figure a1 ) .
this scenario requires that the quasar s far - uv continuum flux changed by a factor of roughly two in @xmath40.63 yr ( rest ) .
little is known about the far - uv variability properties of luminous quasars .
nonetheless , this type of continuum change seems plausible given that 1 ) near - uv flux variations as large as 30 - 40% have been observed in other luminous quasars on times scales of months ( rest , * ? ? ?
* ) , and 2 ) flux changes in the far - uv could be much larger than the near - uv because shorter continuum wavelengths generally exhibit larger amplitude fluctuations @xcite .
the recent sdss stripe 82 survey shows specifically that the near - uv continuum variability of j2123 - 0050 is typical of luminous quasars @xcite .
this survey includes roughly 60 epochs of imaging photometry collected over a @xmath87 year period that overlaps with our spectroscopic observations .
examination of these data for j2123 - 0050 ( kindly provided by kasper schmidt ) shows continuum variations up to @xmath80.27 magnitudes ( or @xmath828% ) on observed time scales of @xmath82 years in the @xmath113 band ( @xmath82000 rest ) .
moreover , the @xmath113-band flux increased by this amount between roughly 2005.8 and the last stripe 82 data point in 2008.0 , which is consistent with an our conclusion that the ionization increased between the keck and vlt observations in 2006.64 and 2008.67 .
the main caveat to interpreting the nal variabilities in terms far - uv flux changes is that there _ might _ have been a large nal variation between 2006.64 ( keck ) and 2006.71 ( mmt ) , i.e. , in just @xmath88 days in the quasar frame ( figure 3 ) .
if this extreme short - term variability can be confirmed , it might be difficult to explain via factor of @xmath82 changes in the far - uv emission from the accretion disk ( but see also * ? ? ?
the outflow nals in j2123 - 005 require at least five distinct absorbing structures that have similar physical conditions ( ionization , column densities , covering fractions , 3.3 and 3.4 ) and roughly similar kinematics ( lines dispersions and velocity shifts , 3.5 ) .
the partial covering evident in all five systems indicates that the absorbers are each characteristically smaller than the quasar s uv continuum source , e.g. , @xmath80.01 - 0.02 pc across .
all five absorbers must also be spatially inhomogeneous to explain the observed changes in the covering fractions if those changes are , indeed , caused by changes in the quasar s ionizing flux ( 4.3 ) .
the absorbing structures might be discrete blobs or filaments created by instabilities in the inner flow , perhaps resembling the clumpy structures seen in some numerical simulations of bal winds @xcite .
if this assessment is correct , then their most likely location is near the minimum radius of @xmath114 pc inferred from the lack of acceleration ( 3.5 ) .
this is because small blobs or filaments have short survival times against dissipation if there is no external pressure confinement .
the characteristic absorber diameter of @xmath115 pc ( 3.4 ) combined with the doppler parameters @xmath116 km s@xmath1 of the broader systems d and e ( table 3 ) indicate dissipation times of order @xmath117 yr .
this is much longer than our observation timeline ( table 1 ) and therefore not a factor in the line variability . however , gas blobs moving radially at @xmath118 km s@xmath1 ( for systems d and e ) will travel a distance of just @xmath83 pc before dissipating .
therefore , they should be located within @xmath83 pc of their point of origin . if that was the inner flow near the accretion disk ( e.g. , at radii not much greater than a few times @xmath119 pc , 2.1 , * ? ? ?
* ; * ? ? ?
* ) , then the absorbers should reside near the @xmath114 pc minimum radius ( 4.1 ) .
one puzzling constraint on the internal structure of these absorbers comes from the high densities . at @xmath114 pc ,
the densities need to be near the upper end of the estimated range , @xmath120 @xmath79 , to avoid over - ionization by the intense quasar radiation field ( 3.6 ) .
however , this entire density range is much larger than the average gas density , @xmath121 @xmath79 , expected for absorbing clouds with total column density @xmath122 @xmath37 ( 3.6 ) and characteristic size @xmath80.015 pc .
this disparity implies that each of the five absorbers is composed of dense sheets or sub - structures that have radial thickness much smaller than their characteristic _ transverse _ size of @xmath80.015 pc .
the combined radial thickness of these sheets or sub - structures in each absorbing region is just @xmath123 cm if they are at @xmath114 pc , or @xmath8@xmath124 cm at @xmath125 pc .
some early studies of bals presented similar arguments for the existence of dense sub - structures with an overall small volume filling factor in quasar outflows @xcite .
those studies also recognized that the creation and survival of such small sub - structures present a serious theoretical challenge for outflow models @xcite .
these concerns about sub - structure effectively ended when @xcite and @xcite proposed an attractive alternative , namely , that the flows could be spatially smooth and continuous if there is a shielding medium at the base of the flow that blocks most of the quasar s intense x - ray and far - uv radiation . with this shielding medium in place , the outflow gas behind the shield should be able to maintain the observed moderate degrees of ionization at much lower gas densities , thus allowing the flow to have an overall smooth spatial distribution .
subsequent x - ray observations supported this scenario by showing that bal quasars are heavily absorbed in x - rays @xcite .
however , j2123 - 0050 , like other nal and mini - bal outflow quasars , does not have strong x - ray absorption ( 1 and 2.2 ) and therefore it does not have a significant x - ray / far - uv shield ( see also the calculations in * ? ? ?
small dense sub - structures do appear necessary to explain the moderate degree of ionization at the inferred flow locations .
thus the debate about the existence of small sub - structures in quasar outflows remains open .
these sub - structures might resemble the small - scale clumps identified recently in one detailed observational study of partial covering in a bal outflow @xcite .
the most remarkable aspect of the nal outflow in j2123 - 0050 is the degree to which it contrasts with bal flows .
for example , it has high speeds and ionizations very similar to bals , but the nals we measure are about a hundred times narrower and a hundred times weaker ( in rew ) than typical bals ( c.f . ,
table 2 and * ? ? ?
* ; * ? ? ?
our estimate of the total column density , @xmath126 @xmath37 , is also 1 - 2 orders of magnitude less than the best available estimates obtained from bals @xcite .
the weak or absent x - ray absorption in j2123 - 0050 @xcite is typical of other nal and mini - bal outflow sources @xcite , but it too is very different from the strong x - ray absorption found in bal quasars @xcite .
this disparity in the x - ray properties between bal , mini - bal and nal outflows has important implications for outflow physics . as noted in 4.4
, the prevailing models of radiatively - driven bal outflows have a shielding medium at the base of the flow that can absorb the quasar s ionizing far - uv / x - ray flux and thus prevent the over - ionization of the outflow gas behind the shield @xcite .
moderate ionizations in the outflow gas are essential to maintain significant opacities and facilitate the radiative acceleration to high speeds . however , the absence of strong x - ray absorption in j2123 - 0050 shows that high speeds and moderate ionizations can occur in quasar outflows _ without _ significant radiative shielding .
one way to reconcile these results is by a unified scheme where the outflows measured via bals , nals and mini - bals coexist at different spatial locations . in situations like this ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , the acceleration of the nal and mini - bal gas might occur in the shielded bal environment ( near the accretion disk plane ) before it moves to its observed location ( farther above the disk ) where the shielding is negligible .
the discrete appearance of nals and mini - bals in quasra spectra might be due to their formation in discrete blobs or filaments that are created via instabilities along the ragged edge of the main bal outflow @xcite .
however , it is not clear there can be enough vertical force to push these blobs significantly away from the disk plane and out of their natural radial trajectories away from the uv continuum source .
a more basic problem with this picture is that nals and mini - bals have degrees of ionization similar to bals even though they are not behind the radiative shield that is alleged to be important for controlling the outflow ionization . if the nal and mini - bal gas is accelerated in a shielded bal environment where it has moderate bal - like ionizations , how does it maintain those moderate bal - like ionizations after it emerges from the shielded bal zone ?
the obvious answer from our analysis of j2123 - 0050 ( 3.6 and 4.4 ) is that the outflow ionization is regulated mostly by locally high gas densities and not by a radiative shield .
high densities could keep the ionization low enough for radiative acceleration even if there is no shielding at all .
perhaps this is the case generally in quasar outflows , including the bals .
however , high densities also lead to the uncomfortable conclusion that the outflows are composed of small sub - structures that have an overall small volume filling factor ( 4.4 ) .
the physical processes that might create or maintain these sub - structures are not understood @xcite .
nonetheless , the evidence for their existence in j2123 - 0050 is at least consistent with high densities playing a more important role in the acceleration physics than radiative shielding .
here we estimate the total mass and kinetic energy in the j2123 - 0050 outflow to determine if it can plausibly be important for feedback to the host galaxy s evolution ( 1 ) .
if the flow geometry is approximately like part of a thin spherical shell , then its total mass is given by @xmath127 where @xmath128 @xmath37 is our best guess at the total column density in all five absorbing systems ( 3.6 ) , @xmath129 pc is a likely radial distance ( 4.1 and 4.4 ) , and @xmath130 is the global covering fraction of the outflow , i.e. , the fraction of 4@xmath131 steradians covered by the flow as seen from the central continuum source @xcite .
the value of @xmath130 is not constrained by our data , but @xmath132 is a reasonable guess based on estimates of the detection rates of high - velocity nal outflows ( 4.2 ) .
the outflow kinetic energy defined by @xmath133 is therefore @xmath134 where @xmath135 km s@xmath1 is roughly representative of all five systems . to estimate the mass loss rate and kinetic energy luminosity
, we would need to make additional assumptions about the duration and radial structure of the flow @xcite .
however , if we simply divide @xmath136 from equation 5 by a characteristic flow time , @xmath137 yr ( for @xmath114 pc and @xmath138 km s@xmath1 ) , we obtain a time - averaged kinetic luminosity of @xmath139 ergs s@xmath1 and a ratio relative the quasar s bolometric photon luminosity of only @xmath140 ( 2.1 ) .
this estimate is much less than the ratio of @xmath141% expected if the outflow is to be important for feedback @xcite .
we also note that the total kinetic energy given by equation 5 is equivalent to the non - neutrino output of only @xmath820 type ii supernovae ( at @xmath142 ergs each )
. the total mass and kinetic energy could be much larger if the outflow gas is near the maximum radius , @xmath125 pc , derived in 3.6 . in that case ,
equations 4 and 5 indicate @xmath143 m@xmath144 and @xmath145 ergs , respectively .
however , this value of @xmath136 is still far less than the binding energy of gas in a massive galaxy ( @xmath8@xmath146-@xmath147 ergs ) , and the increased flow time at @xmath125 pc leads to a time - averaged kinetic energy yield , @xmath148 , that is still two orders of magnitude below the fiducial mark of @xmath141% needed for feedback . therefore , at any of the plausible absorber locations , the kinetic energy yield from this nal outflow is much too small to be important for feedback to the host galaxy s evolution .
we discuss multi - epoch spectroscopic observations of five outflow nal systems in the luminous quasar j2123 - 0050 .
the lines have velocity shifts from @xmath09710 to @xmath014,050 km s@xmath1 and line widths in the range fwhm @xmath8 62 to 164 km s@xmath1 ( as measured in 2006.64 , table 2 ) .
these are the highest velocity nals reported so far to have their formation in a quasar outflow confirmed by all three indicators of line variability , partial covering of the quasar continuum source , and smooth super - thermal line profiles .
the five distinct absorption line systems require five outflow structures with similar physical conditions , similar sizes ( covering fractions ) and roughly similar kinematics .
all five systems have stronger absorption in than , while lower ions such as , and are significantly absent .
the observed nal variabilities were well coordinated between the five systems , with changes in the line strengths accompanied by nearly commensurate changes in the line - of - sight covering fractions .
the time scale for significant nal changes is @xmath40.63 yr in the quasar rest frame .
our analysis of these data provides the following additional constraints on the outflow properties : \1 ) _ absorbing region sizes . _
partial covering of the uv continuum source indicates that the and absorbing regions have characteristic sizes of order @xmath80.01 - 0.02 pc ( 3.4 ) .
\2 ) _ ionization . _
the absence of lines and stronger absorption in compared to in all five systems indicates an ionization parameter of @xmath77 ( 3.6 ) .
the most likely value is near @xmath149 based on the different variabilities observed in and ( 4.3 ) .
\3 ) _ variability . _
the coordinated nature of the nal variations occurring in @xmath40.63 yr is best explained by global ionization changes in the outflow caused by changes in the quasar s ionizing flux ( 4.3 ) .
the magnitude of the ionization changes corresponds to roughly 0.2 - 0.3 dex in @xmath150 .
\4 ) _ metallicity . _
the secure detection of in one system indicates a metal abundance of [ c / h ] @xmath151 , with a lower limit independent of ionization uncertainties of [ c / h ] @xmath96 ( 3.7 ) .
these results are consistent with other estimates of quasar metallicities and they support the claim that the variable nals in j2123 - 0050 form in a quasar outflow .
\5 ) _ total column density .
_ we estimate the total hydrogen column density in system c to be @xmath75 @xmath37 based on @xmath149 and the strength of in that system . from this and the many similarities between the variable systems , we infer that the total column density in all five systems is very roughly @xmath76 @xmath37 ( 3.6 ) .
\6 ) _ dynamics .
_ an upper limit on the radial acceleration , @xmath53 km s@xmath1 yr@xmath1 , and an apparent double line - lock in the systems suggest that i ) the outflow was radiatively accelerated , ii ) it is now gravitationally unbound and coasting freely beyond the main acceleration zone , and iii ) its trajectory is within @xmath816@xmath9 of the purely radial ( line - of - sight ) direction ( 3.5 ) .
\7 ) _ location .
_ the acceleration upper limit suggests that the outflow lines form in a low gravity environment @xmath65 pc from the central quasar ( 3.5 ) .
the minimum gas density needed for recombination in @xmath40.65 yr sets an upper limit on the distance of @xmath81100 pc ( in photoionization equilibrium , 3.6 ) .
thus the full range of plausible distances is @xmath7 pc .
if the absorbing structures are blobs or filaments created in the inner flow , their short survival times against dissipation ( in the absence of a confining pressure ) indicates that they are near the minimum radius of @xmath85 pc ( 4.4 ) .
\8 ) _ density & sub - structure . _ the gas densities corresponding to distances @xmath7 pc are @xmath120 @xmath79 for a gas in photoionization equilibrium at @xmath69 ( 3.6 ) .
these densities are much larger than the average value @xmath121 @xmath79 expected from the total column density and characteristic size .
therefore , each of the five absorbers is composed of dense thin sheets or sub - structures with an overall small volume filling factor ( 4.4 ) .
\9 ) _ radiative shielding & acceleration . _
the absence of strong x - ray absorption in j2123 - 0050 @xcite is typical of nal and mini - bal outflows but contrasts markedly with bal quasars .
it implies that radiative shielding in the x - ray / far - uv is not needed to keep the outflow ionizations moderate ( and bal - like ) and therefore , perhaps , it is not needed to facilitate the radiative acceleration of the flow to high speeds .
we argue that the ionization is controlled , instead , by locally high gas densities in small outflow sub - structures ( 4.5 ) .
\10 ) _ energetics & feedback . _ at the most likely radial distance near @xmath114 pc , this outflow has a total mass of only @xmath814 m@xmath144 and kinetic energy @xmath8@xmath152 ergs . at any of the plausible larger distances noted above ,
the kinetic energy yield from this outflow is still at least two orders of magnitude too small to be important for feedback to the host galaxy s evolution ( 4.6 ) .
we are grateful to dan capellupo , anton koekemoer and leah simon for helpful discussions , and to kasper schmidt for providing the sdss stripe 82 survey data for our analysis in 4.3 .
we thank an anonymous referee for helpful comments on the manuscript .
this work was based in part on observations at the w. m. keck observatory , which is operated as a partnership between the university of california , the california institute of technology and the national aeronautics and space administration , and at eso vlt at the paranal observatories under programme i d 081.a-0242 .
fh acknowledges support from the chandra award tm9 - 0005x .
nk acknowledges support from the department of science and technology through a ramanujan fellowship .
jxp is supported in part by an nsf career grant ( ast-0548180 ) and by nsf grant ast-0908910 .
mtm thanks the australian research council for a qeii research fellowship ( dp0877998 ) .
wu acknowledges financial support from the netherlands foundation for fundamental research of matter ( fom ) .
j. k. , et al .
2008 , apjs , 175 , 297 n. , et al .
2001 , apj , 561 , 118 n. , et al .
2007 , apj , 658 , 829 n. , kaastra j. , kriss g. a. , korista k. t. , gabel j. , proga d. , 2005 , apj , 620 , 665 n. , moe m. , costantini e. , korista k. t. , benn c. , ellison s. , 2008 , apj , 681 , 954 m. , rothenflug r. , 1985 , a&as , 60 , 425 m. , grevesse n. , sauval a. j. , scott p. , 2009
, ara&a , 47 , 481 t. a. , sargent w. l. w. , 1997 , aj , 113 , 136 r. h. , white r. l. , helfand d. j. , 1995 , apj , 450 , 559 m. c. , denney k. d. , peterson b. m. , pogge r. w. , 2007 , in ho l. c. , wang j .- w . , eds , the central engine of active galactic nuclei vol . 373 of astronomical society of the pacific conference series , refining the radius - luminosity relationship for active galactic nuclei , p. 380
e. , milgrom m. , 1989 , apj , 342 , 100 d. , hamann f. rodrguez hidalgo p. , shields j. , 2010 , mnras submitted g. , charlton j. , eracleous m. , giustini m. , hidalgo p. r. , ganguly r. , hamann f. , misawa t. , tytler d. , 2009 , new astronomy review , 53 , 128 m. , 1997 , in arav n. , shlosman i. , weymann r. j. , eds , mass ejection from active galactic nuclei vol .
128 of astronomical society of the pacific conference series , an overview of dynamical models for outflows in bal qsos and seyferts , p 233 m. , korista k. t. , arav n. , 2002 , apj , 580 , 54 t. , springel v. , hernquist l. , 2005 , nat , 433 , 604 m. , hamann f. , shields j. c. , constantin a. , heidt j. , jger k. , vestergaard m. , wagner s. j. , 2003 , apj , 589 , 722 v. , cristiani s. , romano d. , granato g. l. , danese l. , 2004 , mnras , 351 , 976 m. , et al .
1994 , apjs , 95 , 1 j. e. , 2005 , apj , 631 , 689 d. , lacy m. , priddey r. , borys c. , afonso j. , 2007 , apj , 662 , l59 g. j. , korista k. t. , verner d. a. , ferguson j. w. , kingdon j. b. , verner e. m. , 1998 , pasp , 110 , 761 c. b. , weymann r. j. , morris s. l. , turnshek d. a. , 1987 , apj , 317 , 450 c. b. , weymann r. j. , peterson b. m. , sun l. , malkan m. a. , chaffee jr . f. h. , 1986 , apj , 307 , 504 j. r. , et al .
2005 , apj , 623 , 85 j. r. , arav n. , kim t .- s .
, 2006 , apj , 646 , 742 s. c. , brandt w. n. , chartas g. , garmire g. p. , 2002
, apj , 567 , 37 s. c. , brandt w. n. , chartas g. , priddey r. , garmire g. p. , sambruna r. m. , 2006 , apj , 644 , 709 r. , bond n. a. , charlton j. c. , eracleous m. , brandt w. n. , churchill c. w. , 2001 , apj , 549 , 133 r. , brotherton m. s. , 2008 , apj , 672 , 102 r. , eracleous m. , charlton j. c. , churchill c. w. , 1999 , aj , 117 , 2594 r. , masiero j. , charlton j. c. , sembach k. r. , 2003 , apj , 598 , 922 c. m. , 2008 , in revista mexicana de astronomia y astrofisica conference series vol .
32 of revista mexicana de astronomia y astrofisica conference series , accretion disks and the nature and origin of agn continuum variability , p. 1
r. r. , brandt w. n. , gallagher s. c. , hewett p. c. , schneider d. p. , 2010
, apj , 713 , 220 r. r. , brandt w. n. , gallagher s. c. , schneider d. p. , 2009 ,
apj , 696 , 924 r. r. , et al .
2009 , apj , 692 , 758 r. r. , brandt w. n. , schneider d. p. , gallagher s. c. , 2008 , apj , 675 , 985 g. l. , de zotti g. , silva l. , bressan a. , danese l. , 2004 , apj , 600 , 580 p. j. , mathur s. , 1996 , apj , 462 , 637 p. b. , hutsemkers d. , anderson s. f. , brinkmann j. , fan x. , schneider d. p. , york d. g. , 2003 , apj , 593 , 189 p. b. , sadavoy s. i. , hutsemekers d. , everett j. e. , rafiee a. , 2007 , apj , 665 , 174 f. , 1997 , apjs , 109 , 279 f. , 1998 , apj , 500 , 798 f. , 2000 , quasistellar objects : intrinsic agn absorption lines f. , barlow t. , cohen r. d. , junkkarinen v. , burbidge e. m. , 1997 , in arav n. , shlosman i. , weymann r. j. , eds , mass ejection from active galactic nuclei vol .
128 of astronomical society of the pacific conference series , high - velocity narrow - line absorbers in qsos , p. 19
f. , barlow t. a. , beaver e. a. , burbidge e. m. , cohen r. d. , junkkarinen v. , lyons r. , 1995 , apj , 443 , 606 f. , barlow t. a. , junkkarinen v. , 1997 , apj , 478 , 87 f. , barlow t. a. , junkkarinen v. , burbidge e. m. , 1997 , apj , 478 , 80 f. , ferland g. , 1999 , ara&a , 37 , 487 f. , kaplan k. f. , hidalgo p. r. , prochaska j. x. , herbert - fort s. , 2008 , mnras , 391 , l39 f. , korista k. t. , ferland g. j. , warner c. , baldwin j. , 2002 , apj , 564 , 592 f. , korista k. t. , morris s. l. , 1993 , apj , 415 , 541 f. , sabra b. , 2004 , in richards g. t. , hall p. b. , eds , agn physics with the sloan digital sky survey vol .
311 of astronomical society of the pacific conference series , the diverse nature of intrinsic absorbers in agns , p. 203
f. , simon l. e. , 2010 , submitted to mnras f. w. , barlow t. a. , chaffee f. c. , foltz c. b. , weymann r. j. , 2001 , apj , 550 , 142 n. , rix h. , 2004 , apj , 604 , l89
s. , prochaska j. x. , dessauges - zavadsky m. , ellison s. l. , howk j. c. , wolfe a. m. , prochter g. e. , 2006 , pasp , 118 , 1077 p. c. , wild v. , 2010 , mnras , 405 , 2302 p. f. , elvis m. , 2010 , mnras , 401 , 7 p. f. , richards g. t. , hernquist l. , 2007 , apj , 654 , 731 d. w. , brandt w. n. , shemmer o. , steffen a. t. , schneider d. p. , chartas g. , garmire g. p. , 2007
, apj , 665 , 1004 k. f. , et al .
2010 , pasp , in press s. , brandt w. n. , maoz d. , netzer h. , schneider d. p. , shemmer o. , 2007 , apj , 659 , 997 g. , haehnelt m. , 2000 , mnras , 311 , 576 k. t. , voit g. m. , morris s. l. , weymann r. j. , 1993 , apjs , 88 , 357 j. h. , horne k. , kallman t. r. , malkan m. a. , edelson r. a. , kriss g. a. , 1991 , apj , 371 , 541 k. m. , hamann f. , casebeer d. a. , grupe d. , 2009 , apj , 701 , 176 a. l. , et al .
2010 , mnras , 403 , 1541 a. , hunt l. k. , 2003 , apj , 589 , l21 s. , et al .
2000 , apj , 533 , l79 n. , s. e. , prochaska j. x. , tumlinson j. , 2010 , in prep .
t. , charlton j. c. , eracleous m. , ganguly r. , tytler d. , kirkman d. , suzuki n. , lubin d. , 2007 , apjs , 171 , 1 t. , eracleous m. , charlton j. c. , kashikawa n. , 2007 , apj , 660 , 152 t. , eracleous m. , charlton j. c. , tajitsu a. , 2005 , apj , 629 , 115 t. , eracleous m. , chartas g. , charlton j. c. , 2008 , apj , 677 , 863 n. , chiang j. , 1997 , apj , 474 , 91 n. , chiang j. , grossman s. a. , voit g. m. , 1995 , apj , 451 , 498 t. , marconi a. , maiolino r. , 2006 , a&a , 447 , 157 d. , hamann f. , barlow t. , burbidge e. m. , cohen r. d. , junkkarinen v. , lyons r. , 2004 , apj , 601 , 715 d. , hamann f. , rodriguez hidalgo p. , 2008
, mnras , 386 , 2055 h. , lira p. , trakhtenbrot b. , shemmer o. , cury i. , 2007 , apj , 671 , 1256 b. m. , 1997 , an introduction to active galactic nuclei j. x. , hennawi j. f. , 2009 , apj , 690 , 1558 j. x. , omeara j. m. , herbert - fort s. , burles s. , prochter g. e. , bernstein r. a. , 2006 , apj , 648 , l97 g. e. , prochaska j. x. , omeara j. m. , burles s. , bernstein r. a. , 2010 , apj , 708 , 1221 d. , kallman t. r. , 2004 , apj , 616 , 688 d. , stone j. m. , kallman t. r. , 2000 , apj , 543 , 686 g. t. , 2001 , apjs , 133 , 53 g. t. , et al .
2006 , apjs , 166 , 470 p. , hamann f. , nestor d. b. , shields j. , 2010 , in prep . d. s. , veilleux s. , sanders d. b. , 2005 , apjs , 160 , 115 b. d. , sembach k. r. , 1991 , apj , 379 , 245 e. , oh s. p. , 2004
, apj , 608 , 62 k. b. , marshall p. j. , rix h. , jester s. , hennawi j. f. , dobler g. , 2010 , apj , 714 , 1194 o. , netzer h. , maiolino r. , oliva e. , croom s. , corbett e. , di fabrizio l. , 2004 , apj , 614 , 547 y. , et al .
2007 , aj , 133 , 2222 l. e. , hamann f. , 2010 , submitted to mnras r. , petitjean p. , 2000 , a&a , 357 , 414 r. , petitjean p. , ledoux c. , hazard c. , 2002 , mnras , 336 , 753 a. t. , strateva i. , brandt w. n. , alexander d. m. , koekemoer a. m. , lehmer b. d. , schneider d. p. , vignali c. , 2006 , aj , 131 , 2826 i. v. , brandt w. n. , schneider d. p. , vanden berk d. g. , vignali c. , 2005 , aj , 130 , 387 r. c. , zheng w. , kriss g. a. , davidsen a. f. , 2002 , apj , 565 , 773 s. , et al .
2002 , apj , 574 , 740 j. r. , et al .
2006 , apjs , 165 , 1 d. a. , 1995 , in g. meylan ed . , qso absorption lines the covering factors , ionization structure , and chemical composition of qso broad absorption line region gas , p. 223
d. , fan x .-
, 1992 , apjs , 79 , 1 d. e. , et al .
2004 , apj , 601 , 692 d. a. , verner e. m. , ferland g. j. , 1996 , atomic data and nuclear data tables , 64 , 1 m. , peterson b. m. , 2006 , apj , 641 , 689 c. , hamann f. , dietrich m. , 2004 , apj , 608 , 136 r. j. , morris s. l. , foltz c. b. , hewett p. c. , 1991 , apj , 373 , 23 r. j. , turnshek d. a. , christiansen w. a. , 1985 , in j. s. miller ed .
, astrophysics of active galaxies and quasi - stellar objects broad absorption line quasars ( balqsos ) , p. 333
r. j. , williams r. e. , peterson b. m. , turnshek d. a. , 1979 , apj , 234 , 33 v. , et al .
2008 , mnras , 388 , 227
we use the numerical code cloudy , version 08.00 ( last described by * ? ? ?
* ) , to compute ionization fractions for clouds that are in photoionization equilibrium with a quasar continuum source .
the results have wide applicability to quasar outflows and a variety of other near - quasar environments .
the calculations assume the gas clouds are dust free , at a constant density throughout , optically thin in the lyman continuum and geometrically thin such that their radial thickness , @xmath153 , is small compared to their distance , @xmath154 , from the ionizing emission source , i.e. , @xmath155 .
the optically thin assumption ensures that the model clouds have a uniform ionization structure throughout .
this is appropriate for the majority of quasar outflow systems , including j2123 - 0050 , that are highly ionized with @xmath156 @xmath37 .
( moderate continuum opacities can occur at the ionization edge for @xmath157 @xmath37 , but this has no significant affect on the ionization fractions discussed here , * ? ? ?
* ) the calculations assume that the metallicity is solar and the total hydrogen density is @xmath158 @xmath79 , although the specific values of these parameters do not affect the ionization results @xcite .
the quasar spectrum is a simple segmented powerlaw consistent with observational constraints across the most important far - uv energies ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and refs .
therein ) .
it has the form @xmath159 with slope @xmath160 from 0.5 ev to 13.6 ev , @xmath161 from 13.6 ev to 136 ev , and @xmath160 from 136 ev to 136 kev .
sharper cutoffs are applied to the high and low energy extremes ( @xmath162 below 0.5 ev and @xmath163 above 136 kev ) that have no affect on the computed ionizations .
this spectrum yields @xmath164 ( appropriate for luminous radio - quiet quasars , * ? ? ?
* ; * ? ? ?
* ) , and a bolometric correction factor at 1450 of @xmath165 ( see also * ? ? ?
we describe the ionization state of the gas in terms of the dimensionless ionization parameter , @xmath166 , where @xmath167 is the hydrogen - ionizing photon luminosity of the central source @xcite . for the continuum shape described above
, this yields the following expression for the radial distance @xmath168 figure a1 shows the calculated ionization fractions for and several metal ions commonly observed in near - uv spectra of quasars , namely , ions of c , n , o and si . one important application of these results to absorption line work is derivations of metal - to - hydrogen ( m / h ) abundance ratios based on this relation @xmath169 \ = \
\log\left({{n({\rm m}_i)}\over{n({\mbox{h\,{\sc i}}})}}\right ) + \log\left({{f({\mbox{h\,{\sc i}}})}\over{f({\rm m}_i)}}\right ) + \log\left({{\rm h}\over{\rm m}}\right)_{\odot}\ ] ] where ( h / m)@xmath144 is the solar abundance ratio by number , and @xmath170 and @xmath171 are the column densities and ionization fractions , respectively , for and the metal m in ion stage @xmath172 .
ideally , one would apply equation 2 after comparing the measured column densities in several metal ions to the calculated results ( figure a1 ) in order to define @xmath150 and an appropriate ionization correction factor , @xmath173 .
however , if ionization constraints from the data are not available or not reliable , one can also use minimum values of the ionization corrections in equation a2 to derive firm lower limits on the metal - to - hydrogen abundance ratios @xcite .
table a1 lists the minimum correction factors for various metal ions , @xmath174 , together with the values of the ionization parameter , @xmath71 , where these minima occur .
this information is provided in the table according to the element ( column 1 ) and ionization stage ii - vi ( columns 2 - 6 ) for ions that are sometimes detected in quasar outflow spectra .
note that the minimum correction factors occur near , but at larger @xmath150 than , the peaks in the @xmath175 curves shown in figure a1 . for convenience
, we note that the most recent solar abundance ratios are @xmath176h / m@xmath177 for carbon , 4.17 for nitrogen , 3.31 for oxygen , 5.55 for aluminum , 4.49 for silicon , 6.59 for phosphorus and 4.88 for sulfur @xcite . & + carbon & @xmath01.75 & @xmath03.32 & @xmath03.80 & @xmath05.39 & + nitrogen & @xmath01.75 & @xmath03.28 & @xmath04.15 & @xmath04.55 & @xmath05.89 + oxygen & @xmath01.81 & @xmath03.02 & @xmath04.17 & @xmath04.69 & @xmath05.06 + aluminum & @xmath02.12 & @xmath02.20 & @xmath03.36 & @xmath04.11 & @xmath04.96 + silicon & @xmath01.65 & @xmath02.93 & @xmath03.03 & @xmath03.89 & @xmath04.51 + phosphorus & @xmath01.90 & @xmath03.10 & @xmath03.34 & @xmath03.57 & @xmath04.42 + sulfur & @xmath01.50 & @xmath02.95 & @xmath03.73 & @xmath03.73 & @xmath03.97 + + & + carbon & @xmath02.60 & @xmath01.55 & @xmath00.75 & 0.30 & + nitrogen & @xmath03.20 & @xmath01.50 & @xmath00.75 & @xmath00.15 & 0.65 + oxygen & @xmath03.15 & @xmath01.65 & @xmath00.75 & @xmath00.15 & 0.30 + aluminum & @xmath02.60 & @xmath02.10 & @xmath01.50 & @xmath00.70 & @xmath00.10 + silicon & @xmath02.80 & @xmath01.95 & @xmath01.55 & @xmath01.00 & @xmath00.35 + phosphorus & @xmath02.60 & @xmath01.80 & @xmath01.35 & @xmath01.00 & @xmath00.50 + sulfur & @xmath02.90 & @xmath01.65 & @xmath01.20 & @xmath00.85 & @xmath00.60 + | we report on the discovery of a high - velocity narrow absorption line outflow in the redshift 2.3 quasar j212329.46 - 005052.9 .
five distinct outflow systems are detected with velocity shifts from @xmath09710 to @xmath014,050 km s@xmath1 and @xmath2@xmath21548,1551 line widths of fwhm @xmath3 62 to 164 km s@xmath1 .
this outflow is remarkable for having high speeds and a degree of ionization similar to broad absorption line ( bal ) flows , but line widths roughly 100 times narrower than bals and no apparent x - ray absorption .
this is also , to our knowledge , the highest - velocity narrow absorption line system confirmed to be in a quasar outflow by all three indicators of line variability , smooth super - thermal line profiles and doublet ratios that require partial covering of the quasar continuum source .
all five systems have stronger absorption in @xmath2@xmath21032,1038 than with no lower ionization metal lines detected .
their line variabilities also appear coordinated , with each system showing larger changes in than and line strength variations accompanied by nearly commensurate changes in the absorber covering fractions .
the metallicity is approximately twice solar .
these data require five distinct outflow structures with similar kinematics , physical conditions and characteristic sizes of order 0.01 - 0.02 pc ( based on partial covering ) .
the coordinated line variations , occurring on time scales @xmath40.63 yr ( quasar frame ) , are best explained by global changes in the outflow ionization caused by changes in the quasar s ionizing flux
. an upper limit on the acceleration , @xmath53 km s@xmath1 yr@xmath1 , is consistent with blobs of gas that are gravitationally unbound and coasting freely @xmath65 pc from the central black hole .
additional constraints from the variability time indicate that the full range of plausible distances is @xmath7 pc .
however , if these small absorbing structures were created in the inner flow , they should be near the @xmath85 pc minimum radius because they can travel just a few pc before dissipating ( without external confinement ) .
an apparent double line - lock in suggests that the flow was radiatively accelerated and its present trajectory is within @xmath816@xmath9 of the radial ( line - of - sight ) direction .
the absence of strong x - ray absorption shows that radiative shielding in the far - uv and x - rays is not needed to maintain moderate bal - like ionizations and therefore , apparently , it is not needed to facilitate the radiative acceleration to high speeds .
we argue that the ionization is moderated , instead , by high gas densities in small outflow sub - structures .
finally , we estimate that the kinetic energy yield from this outflow is at least two orders of magnitude too low to be important for feedback to the host galaxy s evolution .
[ firstpage ] galaxies : active quasars : general quasars : absorption lines quasars : individual : j212329.46 - 005052.9 |
with the wide stream of new high - quality data flowing from rhic , as well as with the continued efforts at sps ( for recent results see , _
e.g. _ , @xcite ) , there is a growing need for a simple description of the basic underlying physics .
only then our understanding of the phenomena occurring in ultra - relativistic heavy - ion collisions can be pushed forward , and space made for potential new phenomena , hitherto unexplained within the basic picture . in this lecture
we argue that most of the `` soft '' features of the data from rhic ( particle ratios , momentum spectra , hbt correlation radii ) can be explained very efficiently within an embarrassingly simple model , which merges the thermal model @xcite with expansion , and incorporates in a complete way the resonances @xcite .
our description uses hadronic degrees of freedom and starts at freeze - out , _
i.e. _ at the point of the space - time evolution of the system where the hadrons cease to interact .
pertinent theoretical questions , such as what had been happening before freeze - out , what had led to the strong expansion of the system , why is the thermal picture successful , what is the nature of hadronization , not to mention the notorious
_ `` was there quark - gluon plasma ? '' _ , will not and can not be addressed in this lecture .
nevertheless , we believe that our studies prepare ground for such questions .
historically , the ideas of the thermal description of a hadronic system go back to the works of koppe @xcite , fermi @xcite , landau @xcite , and hagedorn @xcite . more recently
, many groups have used these ideas in numerous papers in an effort to explain the data from various relativistic heavy - ion experiments , from sis , through ags and sps , to rhic . along the way
, the original picture has been occasionally supplied with extra features , such as the fugacities controlling deviations from chemical equilibrium @xcite , finite volume and van der waals corrections @xcite , or the use of the canonical instead of the grand - canonical ensemble @xcite .
the works of heinz and collaborators @xcite put forward the concept of two freeze - outs . as the system expands and cools , it first passes through the chemical freeze - out point at temperature @xmath1 .
later , the particles can only rescatter elastically , until these processes are switched off at a lower temperature @xmath2 . in an appealing way the distinct freeze - outs explained the need for a higher temperature to reproduce well the particle ratios , and a much lower temperature to describe the slopes of the momentum spectra . in our work
@xcite we have shown that with the complete treatment of resonances , the distinction between the two freeze - outs is not needed , at least for rhic , and one can achieve very good explanation of all `` soft '' features of data assuming one universal freeze - out , @xmath3 we have also dropped , with the ockham razor at hand , all other additions to the most naive thermal approach .
the dropped features may be reconsidered later on , provided there is a well - established phenomenological need , or theoretical argumentation .
the main ingredients of our model are as follows : * there is one freeze - out , as discussed above , at which all the hadrons occupy the available phase space according to the statistical distribution factors .
the scenario with a single freeze - out is natural if the hadronization occurs in such conditions that neither elastic nor inelastic processes are effective .
an example here is the picture of the supercooled plasma of ref .
moreover , the star collaboration @xcite has presented an important argument in favor of very weak rescattering after the chemical freeze - out at rhic , based on the observation of the @xmath4 peaks in the pion - kaon correlations .
this essentially shows that the expansion time between the chemical and thermal freeze - out is shorter than the life - time of the @xmath4 , _
i.e. _ @xmath5 fm/@xmath6 . additionally , the fact that the measured yields of @xmath4 @xcite are reproduced very well within the thermal model @xcite hints to the scenario with a short time between the two possible freeze - outs , as proposed in ref .
thus , approximation ( [ tuniv ] ) is reasonable .
* a crucial feature of our analysis is the _ complete _ treatment of the hadronic states , with all resonances from the particle data table @xcite included in the analysis of both the ratios and the momentum spectra ( cf .
[ fig : therm ] ) .
although the high - lying states are suppressed by the thermal factors , their number increases according to the hagedorn hypothesis @xcite , such that their net effect is important .
for instance , only a quarter of the observed pions at rhic comes from the `` primordial '' pions present at freeze - out , and three quarters are produced via resonance decays .
all decays , two and three body , are implemented in cascades , with the branching ratios taken from the tables . for the @xmath7-spectra
the resonances are also very important , since their decays increase the slope , as if the temperature were effectively lower @xcite .
it has also been found that the inclusion of resonances speeds up the cooling of the system in hydrodynamic calculations @xcite . * whereas expansion of the system does not alter the particle ratios at midrapidity ( provided the system is boost - invariant , which is a good approximation at rhic , see sec .
[ sec : expansion ] ) , it becomes absolutely essential for the @xmath7-spectra .
we model the expansion ( transverse flow ) and the size of the system with two parameters : the proper time at freeze - out , @xmath8 , and the transverse size , @xmath9 . *
the model has altogether four adjustable parameters : two thermal and two geometric , which possess clear physical interpretation .
the two thermal parameters , temperature , @xmath10 , and the baryon chemical potential , @xmath11 , are fixed by the analysis of the ratios of the particle abundances @xcite .
the two geometric parameters are fixed with help of the @xmath0-spectra .
the invariant time @xmath8 controls the overall normalization of the spectra , while the ratio @xmath12 directly influences their slopes .
all data used in the present study are for au+au collisions at @xmath13 gev .
+ @xmath10 [ mev ] & 165@xmath14 & + @xmath15 [ mev ] & 41@xmath16 & + @xmath17 [ mev ] & 9 & + @xmath18 [ mev ] & -1 & + @xmath19 & 0.97 & + + @xmath20 & @xmath21 & .optimal thermal parameters , ratios @xmath22 used for the fit , and further predicted ratios .
the preliminary experimental numbers for @xmath23 @xcite have changed @xcite , and better agreement with the model followed . [ cols= " < , < " , ] to summarize this section ,
we conclude that the successful and economic description of the spectra , as seen from figs .
[ f0 ] , [ fcen ] , and [ f1 ] , provides a strong support for the thermal approach with universal freeze - out in the description of the ultra - relativistic heavy - ion collisions at rhic .
in the presented model the fitted values for the geometric parameters , @xmath8 and @xmath9 , are low , of the order of the size of the colliding nuclei .
this leads to two problems : 1 ) the values of the hbt radii , as shown in sec .
[ sec : hbt ] would be too small compared to the experiment , and 2 ) there would be little time left for the system to develop large transverse flow .
both problems can be solved with the inclusion of the excluded - volume ( van der waals ) corrections .
such effects were realized to be important already in the previous studies of the particle multiplicities in ultra - relativistic heavy - ion collisions @xcite , where they led to a significant dilution of system . in the case of the classical boltzmann statistics , which is a very good approximation for our system @xcite , the excluded volume corrections bring in a factor @xcite @xmath24 into the phase - space integrals , where @xmath25 denotes the pressure , @xmath26 is the excluded volume for the particle of species @xmath27 , , hence the factor of 4 in the definition of @xmath28 . ] and @xmath29 is the density of particles of species @xmath27 . the pressure must be calculated self - consistently from the equation @xmath30 where @xmath31 is the partial pressure of the ideal gas of hadrons of species @xmath27 . for the simplest case where the excluded volumes for all particles are equal , @xmath32 , @xmath33 , the excluded - volume correction ( [ vdw ] ) produces a scale factor common to all particles , which we can denote by @xmath34 .
the formula ( [ dni ] ) becomes @xmath35 the presence of the factor @xmath34 in eq .
( [ dni2 ] ) may be compensated by rescaling @xmath36 and @xmath8 by the factor @xmath37 .
that way , we retain all the previously obtained results for the particle abundances and the momentum spectra . however , now the system is more dilute and larger in size . next , we present an estimate of @xmath37 . with our values of the thermodynamic parameters we have @xmath38mev/@xmath39 , which leads to @xmath40 with @xmath41fm
. values of this order have been typically obtained in other works .
thus , the excluded - volume corrections can increase the size parameters at freeze - out by about 30% , and in consequence the problems 1 ) and 2 ) are alleviated : the geometric parameters become large enough to be reconciled with expansion , and the hbt correlation radii can be properly reproduced , see sec . [
sec : hbt ] .
, @xmath42 , and their ratio , as predicted by the model ( solid line ) and measured by the phenix collaboration . , width=264 ] the transverse hbt radii @xmath43 and @xmath42 ( here we use the bertsch - pratt @xcite parameterization ) measured @xcite at rhic have values very close to those measured at smaller beam energies . only the longitudinal radius , @xmath44 , exhibits a monotonic growth with @xmath45 @xcite .
the weak energy dependence of @xmath43 and @xmath42 has come as a great puzzle , since the rhic beam energy , @xmath13 gev , is almost one order of magnitude larger than the sps energy , @xmath45 = 17 gev , and based on the hydrodynamic calculations one would expect much larger systems to be produced at rhic .
also , a longer life - time of the firecylinder was expected at rhic , which should be reflected in longer emission times of pions , which in turn would result in the ratio @xmath46 much larger than unity @xcite . on the contrary , the experimental measurements indicate that @xmath46 is compatible with unity in the whole range of the studied transverse - momenta of the pion pair ( @xmath47 gev ) .
this fact is another surprise delivered by the analysis of the rhic data for the pion - pion correlations .
we have computed the pion hbt radii in our model .
the calculation is based on the formalism of ref .
@xcite , and is similar to the case of the particle spectra shown in sec .
[ sec : expansion ] .
details will be presented elsewhere @xcite .
the results of an approximate calculation neglecting the hadronic widths are shown in fig .
[ fig : hbt ] , where the hbt radii are plotted as a function of the transverse momentum of the pion pair , @xmath48 .
we note that very reasonable agreement with the data is achieved .
we have used @xmath40 of sec . [
sec : exclude ] .
in particular , the ratio @xmath46 ( independent of the scale factor @xmath37 ) is close to unity .
the @xmath48 dependence of @xmath43 is a bit too flat .
the longitudinal radius , @xmath44 , is sensitive to the cut - off in the rapidity distribution and can not be reliably computed in the present , boost - invariant , model .
the presented results for the hadron production at rhic support the idea that particles are produced thermally , and this is the basic lesson for today .
the simple , economic model with the approximation of a universal freeze - out , simple expansion , and complete treatment of resonances , predicts the particle ratios , the transverse - momentum spectra , and the hbt correlation radii for the pion in agreement with the data .
we note that the thermal approach works noticeably better at the rhic energies than at lower energies , where , _
e.g. _ , the particle ratios are not described to that accuracy @xcite , or the spectrum of the @xmath49 baryons is not reproduced .
this indicates that the soft physics becomes simpler at rhic , with our model being able to yield the quite impressive results of table 1 and figs .
[ f0 ] , [ fcen ] , [ f1 ] , and [ fig : hbt ] . in our phenomenology the pre - freeze - out
stages are hidden and only the conditions at the moment where the hadrons decouple are relevant .
this provides useful constraints for the more microscopic approaches . these calculations , describing early stages of the evolution ,
should ultimately provide the freeze - out conditions such as , or similar , to the ones used in our study .
certainly , the most challenging theoretical question which remains and should be addressed in future efforts is _ why the model works so nicely _ , and what it means for the underlying physics of particle production and the mechanism of hadronization .
we are grateful professor andrzej budzanowski for his encouragement and interest in this work , to marek gadzicki for numerous helpful discussions , and to boris hippolyte for pointing out the early experimental spectra for the @xmath49 baryons .
for a recent discussion of this issue see : j. rafelski and j. letessier , to appear in the proceedings of pan american advanced studies institute on new states of matter in hadronic interactions ( pasi 2002 ) , campos do jordao , brazil , 7 - 18 jan 2002 , hep - ph/0206145 . | applications of a simple thermal model to ultra - relativistic heavy - ion collisions are presented .
we compute abundances of various hadrons , including particles with strange quarks , the @xmath0-spectra , and the hbt radii for the pion .
surprising agreement is found , showing that the thermal approach can be used successfully to understand and describe the rhic data . |
10 j. putschke , j. phys .
g * 34 * , s679 ( 2007 ) [ arxiv : nucl - ex/0701074 ] .
r. c. hwa and c. b. yang , arxiv:0801.2183 [ nucl - th ] ; c. y. wong , arxiv:0804.4017 [ hep - ph ] .
m. daugherity [ star collaboration ] , pos c * frnc2006 * , 005 ( 2006 ) [ arxiv : nucl - ex/0611032 ] .
s. a. voloshin , phys .
b * 632 * , 490 ( 2006 ) [ arxiv : nucl - th/0312065 ] .
e. v. shuryak , phys .
c * 76 * , 047901 ( 2007 ) [ arxiv:0706.3531 [ nucl - th ] ] . c. a. pruneau , s. gavin and s. a. voloshin , nucl .
a * 802 * , 107 ( 2008 ) .
s. gavin and m. abdel - aziz , phys .
lett . * 97 * , 162302 ( 2006 ) .
j. adams _ et al . _ [ star collaboration ] , j. phys .
g*32 * , l37 , ( 2006 ) .
j. adams _ et al . _ [ star collaboration ] , j. phys .
g*32 * , l37 , ( 2006 ) ; phys .
c * 73 * , 064907 ( 2006 ) .
d. adamova _ et al . _ [ ceres collaboration ] , arxiv:0803.2407 [ nucl - ex ] .
j. adams _ et al . _ [ star collaboration ] , phys . rev .
c * 72 * , 044902 ( 2005 ) [ arxiv : nucl - ex/0504031 ] .
s. gavin , phys .
lett . * 92 * , 162301 ( 2004 ) [ arxiv : nucl - th/0308067 ] .
s. s. adler _ et al .
_ [ phenix collaboration ] , phys .
c * 69 * , 034909 ( 2004 ) . c. adler _ et al .
_ [ star collaboration ] , phys .
c * 66 * , 034904 ( 2002 ) ; j. adams _ et al .
_ , phys .
c * 72 * , 014904 ( 2005 ) . | correlation studies exhibit a ridge - like feature in rapidity and azimuthal angle , with and without a jet trigger .
we ask whether the feature in untriggered correlations can be a consequence of transverse flow and viscous diffusion .
jet correlation measurements display a ridge of enhanced particle emission that is broad in pseudorapidity @xmath0 and narrow in azimuthal angle @xmath1 @xcite .
many have attributed this ` hard ridge ' to the passage of the jet through the high - density environment produced by the nuclear collision ; see e.g. @xcite .
interestingly , a similar ridge feature is exhibited in untriggered correlation studies where no jet tag is present @xcite .
as with the hard ridge , the width of the untagged ` soft ridge ' broadens in @xmath0 and narrows in @xmath1 for collisions of higher centrality ( smaller impact parameter ) .
we ask whether the @xmath2 dependence of the soft ridge in untagged correlations can be a consequence of radial flow and viscous diffusion .
the connection between radial flow and momentum correlations has been stressed elsewhere @xcite . in ref
. @xcite we attributed the rapidity broadening of untriggered momentum correlations to viscous diffusion , extracting an estimate of the viscosity - to - entropy ratio in the range from @xmath3 to @xmath4 . in this paper
we show that viscous diffusion together with flow can explain the centrality dependence of the soft ridge ; see fig . 1 ( right ) . the soft ridge has been observed both in momentum fluctuation studies @xcite and multiplicity correlation measurements @xcite . here , we focus on the momentum correlation results , which we argue depend strongly on the viscosity of the high density liquid produced in collisions .
our hope is to eventually reduce the rather large uncertainty in the viscosity estimate in @xcite by using the full landscape of @xmath5 correlations , rather than just the @xmath0 dependence . in this paper
, we first recall the role of shear viscosity in the evolution of momentum fluctuations .
we survey the hydrodynamic formulation of @xcite , extending it to include transverse flow .
we then confront the soft ridge .
central nuclear collisions produce a fluid that flows outward with an average transverse velocity @xmath6 . in the hydrodynamic description of these collisions
, we typically assume that @xmath6 varies smoothly with spacetime @xmath7 and is the same for all collisions of a fixed impact parameter .
more realistically , small deviations relative to @xmath6 occur throughout the fluid , varying with each collision event .
such deviations occur , e.g. , because the number and location of nucleon - nucleon subcollisions varies in each event .
viscous friction arises as neighboring elements of the fluid flow past one another .
this friction drives the stress - energy tensor toward the local average @xmath8 , where the energy density is @xmath9 , the pressure is @xmath10 , and @xmath11 .
the extent to which variations @xmath12 survive the collision depends on the magnitude of the viscosity and the lifetime of the fluid . near a point where @xmath13 , the momentum current @xmath14 satisfies a diffusion equation , @xmath15 , where the kinematic viscosity is @xmath16 @xcite . to incorporate effects of radial and elliptic flow ,
we assume bjorken longitudinal flow in the @xmath17 direction coupled to a radial flow @xmath18 in the transverse plane that depends only on @xmath19 and @xmath20 . we write
@xmath21 where we take @xmath22 for simplicity .
momentum density fluctuations and their dissipation by viscosity are characterized by the correlation function @xmath23 where the brackets represent an average over an ensemble of events .
the difference @xmath24 of @xmath25 from its local equilibrium value @xmath26 also satisfies a diffusion equation @xcite .
we show in ref .
@xcite that these momentum density correlations are observable by the relation @xmath27 where @xmath28 labels particles from each event and @xmath29 .
the aim is then to reconstruct @xmath24 by measuring @xmath30 in different windows of rapidity and azimuthal angle ; we emphasized the role of rapidity in determining the viscosity in @xcite .
while measurements of @xmath30 would be ideal , much can be learned from @xmath31 correlation measurements at rhic and sps that used somewhat different observables .
star measures the quantity @xmath32 ceres and star also measure the related quantity @xmath33 @xcite .
our covariance @xmath30 is sensitive to the variation of the @xmath31 of particles as well as their number density , since both quantities effect the momentum current density .
in contrast , @xmath34 and @xmath35 are designed to minimize the number density contribution .
these quantities are related by @xmath36 , where @xmath37 measures the number density contribution . .
( right ) centrality dependence of the width of the near side peak in @xmath38 and @xmath39 as a function of centrality .
centrality is determined using @xmath40 as in @xcite.,title="fig:",width=259 ] [ fig:1a ] .
( right ) centrality dependence of the width of the near side peak in @xmath38 and @xmath39 as a function of centrality .
centrality is determined using @xmath40 as in @xcite.,title="fig:",width=288 ] [ fig:1b ] star employs @xmath34 to construct a correlation function as a function of rapidity and azimuthal angle .
in essence , they measure @xmath34 in various @xmath41 and @xmath38 bins , and use the data to construct a correlation function @xmath42 defined by @xmath43 ; see @xcite for details .
they find a near - side ridge peaked at @xmath44 together with other features attributable to momentum conservation , elliptic flow , and perhaps other phenomena .
our primary focus is the near - side peak in azimuth the soft ridge as shown in fig . 1 ( left ) .
figure 1 ( right ) shows the widths of this peak in azimuth @xmath45 and rapidity @xmath46 as functions of centrality @xcite .
centrality is measured using the combination @xmath40 , where @xmath47 and @xmath48 are , respectively , the numbers of binary collisions and participants .
the increase of @xmath46 with centrality was used in @xcite to estimate the shear viscosity .
diffusion increases the width of @xmath25 relative to its value in peripheral collisions by an amount @xmath49 , where @xmath50 and @xmath51 are the freeze out times in central and peripheral collisions .
transverse flow does not appreciably alter @xmath46 , since @xmath6 is only weakly dependent on @xmath0 .
note that the large uncertainty range for the viscosity in @xcite follows from the unknown number density contribution relating @xmath34 to @xmath30 .
we turn now to consider the role of viscous diffusion and transverse flow in determining the width in the azimuthal angle as a function of centrality in fig . 1 ( right ) .
to see why flow is important , we consider a blast wave model of the mean flow . in a central collision , @xmath52 . a fluid cell a distance @xmath53 from the center of the collision volume has a mean speed @xmath6 prior to freeze out
correspondingly , the opening angle into which two particles from this cell emerge is @xmath54 , for a thermal velocity @xmath55 .
this introduces correlations , since particles near the center of the collision volume have a large opening angle , while those from a larger @xmath53 have a smaller @xmath38 .
a similar correlation may affect the hard ridge as well @xcite . to compute the angular correlations that result from flow and diffusion ,
we write the momentum correlation function @xmath56 where @xmath57 is the boltzmann distribution , @xmath58 is the density , and the integral is over the cooper - frye freeze out surface @xmath59 .
solution of the diffusion equation for @xmath60 derived from ( [ eq : temp2 ] ) yields the form @xmath61 where @xmath62 and @xmath63 .
we obtain the azimuthal width @xmath45 by integrating ( [ eq : momcorr ] ) using ( [ eq : spacecorr ] ) .
we use ( [ eq : temp2 ] ) to compute the spatial widths @xmath64 and @xmath65 as functions of the proper time @xmath66 .
the effects of elliptic and radial flow are included by assuming an eccentric blast wave form @xmath67 .
the strength of flow is determined by fitting the measured @xmath68 and elliptic flow coefficient @xmath69 as in fig . 2 . [ fig:1a ] [ fig:1b ] the result shown in fig .
1 ( right ) agrees well with the data .
observe that viscous diffusion and transverse flow have the opposite effect on @xmath45 . in the absence of flow
, @xmath45 would increase as a function of centrality due to diffusion .
consequently , information from @xmath45 and @xmath46 can constrain the viscosity - to - entropy ratio extracted from measurements of @xmath30 .
values in the range @xmath70 are consistent with the data in fig . 2 , with other parameters
held fixed .
the problem remains , however , that @xmath71 is measured , not @xmath30 .
the large uncertainty due to the possible role of density fluctuations remains .
direct measurements of @xmath30 are needed to eliminate that uncertainty ; until such data become available the estimate @xmath72 stands @xcite .
_ note added : _ after the completion of this manuscript we learned of work explaining the amplitude of the soft ridge as a consequence of glasma correlations plus radial flow @xcite .
these authors do not explain the centrality dependence of @xmath45 and @xmath46 , while we do not address the amplitude .
we view these results aa complementary , since glasma may well determine the initial conditions for hydrodynamic evolution .
this work was supported by a u.s .
national science foundation pecase / career award under grant phy-0348559 . |
observations show that the two point correlation function of the galaxies is an approximate power law over a range of scales .
this result , known for decades now , defies a simple first - principle " explanation because of the complexity of the physical processes involved . in the conventional big - bang cosmology
the dominant contribution to the energy density of the universe is in the form of nonbaryonic dark matter and the visible galaxies ( made of baryons ) form due to complex processes of cooling and fragmentation within the dark matter halos .
it is possible to model non linear gravitational clustering of _ dark matter halos _ by a scaling ansatz ( see eg .
@xcite ) and show that a power law correlation is expected ( in a spatial and temporal interval ) for scale invariant initial conditions .
this result arises from the fact that newtonian gravitational dynamics in a @xmath1 universe does not have any preferred scale which is of interest to cosmology . but to translate this result to baryonic structures is not easy and at present the only explanation for the power law correlation of galaxies arises from numerical simulations .
there are , however , alternate ( and less popular ) models for galaxy formation in quasi - steady state cosmology ( qssc ) in which the process is addressed without invoking gravitational instability explicitly @xcite .
it has been claimed , based on numerical simulations , that these models also lead to power law correlation function for galaxy distribution .
this result is of intrinsic interest even to conventional cosmologists who may not accept the alternate model because it provides a scenario to probe the relationship between ( i ) the conventional scenario of galaxy formation via gravitational instability ( accepted by most of the working cosmologists ) and ( ii ) the power law correlation function for the galaxies .
if a completely different model can lead to power law correlation function , then it is clear that observations of galaxy correlation function , by itself , can not be used as a discriminator between the theories .
since the `` rule '' used for creating galaxies in qssc is simple and explicit , it is indeed possible to study this process completely analytically .
we provide such an analysis in this paper and show that the model is intrinsically unstable .
however , the galaxies produced by this rule _ will _ exhibit a power law correlation function for a range of intermediate time scales .
the rest of the paper is arranged as follows . in the next section
, we provide an analytic description of the model for galaxy formation in the qssc , which was used in @xcite .
this is done in terms of discrete time steps to parallel the previous work . in section 3
we discuss the continuum limit of this model that lends itself to a completely tractable analysis . in section 4
we describe a very general class of processes which obey the same equations as described here and a broader class of applications .
the last section gives the conclusions .
a non standard model for galaxy formation in qssc is built along the following lines : consider a set of points in the 3-dimensional space which represents the location of galaxies at a given instant of time .
we now generate a set of new galaxies near each one of the galaxies ( `` non - local self - replication '' ) by a physical process which should quite minimally involve some extra negative energy field .
the key ingredient of the nonstandard cosmological models is the existence of some `` creation field '' so that galaxies could generate new galaxies nearby ( see @xcite ) .
let the probability for any given new galaxy to be located at a distance @xmath2 from an old galaxy be @xmath3 .
this process will increase the number of galaxies as the universe evolves . in qssc ,
a balance between creation and quasi - steady state structure is maintained by rescaling the size of the universe in each cycle .
mathematically , this is taken into account by rescaling each of the 3 dimensions by a factor @xmath4 thereby increasing the volume of available space .
( if we create _
one _ new galaxy near an old one with a probability @xmath5 , then we will be doubling the number of galaxies during the self replication .
the rescaling should be such that the volume doubles giving @xmath6 . for future convenience
, we shall keep @xmath7 as an arbitrary constant . )
we now select a subset of galaxies in the central region such that the total number of galaxies remain the same .
this step renormalizes the process back to the original situation so that the process can now be repeated with the new subset of galaxies . in the specific case of number of galaxies doubling per cycle
, we should take the central region containing half the volume .
it is obvious that such a process of galaxy formation will lead to correlations between the galaxies since new galaxies are created close to the old ones with a probability @xmath5 .
the key question is how the correlation function scales with the distance scale .
we will now provide a mathematical analysis of this problem . the evolutionary rule described above
can be stated mathematically in the form : @xmath8 \label{threeq}\end{aligned}\ ] ] where @xmath9 are constants with @xmath10 and @xmath11 ; @xmath5 is a probability function normalized to unity for integration over all @xmath2 .
we shall also assume that @xmath12 is normalized in such a way that its integral over all @xmath13 is unity .
this gives us the conditions @xmath14 it is obvious that equation ( [ threeq ] ) preserves these conditions under evolution because of the explicit normalization chosen on the right hand side . in the context of the quasi steady state cosmological model
, @xmath12 will be the ratio between the number density of galaxies and the mean density .
the normalization in ( [ twoint ] ) preserves the quasi steady state condition between different cycles under the simultaneous action of matter creation and expansion .
while equation ( [ threeq ] ) was motivated from a particular model for galaxy formation , the rest of the analysis only uses this equation and is independent of the assumptions which go into it . in section 4
we will comment on a wider class of phenomena which could be modeled by such an equation .
let us now consider the solutions to equation ( [ threeq ] ) .
linearity in @xmath12 suggests switching to the fourier space variables @xmath15 with @xmath16 equation ( [ threeq ] ) now reduces to a simple form @xmath17 \label{threek}\ ] ] where we have denoted by the same symbol @xmath18 the fourier transform of the probability function .
given the form of this probability function , this equation iteratively determines the evolution of @xmath19 . while the equation is fairly simple in structure , it is not easy to find its general solution .
note that the conservation condition ( [ twoint ] ) which demands @xmath20 for all @xmath21 is satisfied in this case because @xmath22 for a normalized probability distribution .
this also shows that solutions of the form @xmath23 , with some constant @xmath24 , are unacceptable for @xmath25 because they will violate the normalization condition .
further , since we have the freedom to choose the initial condition @xmath26 , any general solution to equation ( [ threek ] ) must contain one arbitrary function ; it is difficult to obtain such a _ general _ solution .
a special class of solutions to equation ( [ threek ] ) will correspond to a steady state such that @xmath27 .
this function satisfies the equation @xmath28.\label{twof}\ ] ] two trivial solutions to this equation correspond to @xmath29 and , in fact , we shall see later that the most generic initial conditions will drive the system to either of these two limits by our process .
the only nontrivial solution which exists can be found by the iteration in the form @xmath30 \left [ 1 + \lambda w ( { { \bf k } \over \bar\mu } ) \right]}\nonumber \\ & = & f_s \left
( { { \bf k } \over \bar \mu^n}\right ) \prod^n _ { n=0 } { ( 1 + \lambda ) \over 1 + \lambda w ( { \bf k}/ \bar\mu^n ) } .
\end{aligned}\ ] ] taking the logarithms , followed by the limit @xmath31 , and using @xmath32 we get the result @xmath33 .
\label{lnsum}\ ] ] to study the properties of the solution we note that @xmath34 for all @xmath35 .
this is true for a wide class of probability distributions with some characteristic scale @xmath36 , and @xmath37 usually decreases for @xmath38 .
we will also assume that @xmath37 depends only on the magnitude of @xmath39 because of the statistical isotropy of the process ( though our results are easily generalizable to other cases ) .
it follows that each of the denominators in ( [ lnsum ] ) is close to unity for all @xmath40 making all the terms negligibly small for @xmath40 .
so we need to sum the series in ( [ lnsum ] ) only up to @xmath41 $ ] with the integer value , lower than the bound , taken for @xmath42 .
the result of the sum depends on the form of @xmath43 at large @xmath44 .
the simplest case corresponds to assuming the @xmath43 vanishes for @xmath38 which is exact if we take @xmath45 where @xmath46 for @xmath47 and zero otherwise .
then the asymptotic solution is given by @xmath48 ( the equivalence of the two forms follows from simple algebra and definition of @xmath42 . )
this is a power law solution with the index determined essentially by the two parameters of the problem . before proceeding further ,
let us consider the effect of the normalization condition ( [ twoint ] ) more closely .
this is of some interest because one might feel that such a normalization is unnecessarily restrictive and a wider class of phenomena can be modeled by relaxing this condition .
it turns out , however , that our qualitative considerations are not affected by this constraint . relaxing this normalisation
can be done most conveniently by replacing the factor @xmath49 in the denominator of the right hand side of ( [ threeq ] ) by @xmath50 where @xmath51 is a constant different from @xmath52 .
this will change equation ( [ threek ] ) to the form : @xmath53 .
\label{newthreek}\ ] ] setting @xmath54 and using @xmath55 , we get @xmath56 since the integral over all space of @xmath57 is just @xmath58 we find that @xmath59 this shows how the total number of galaxies changes with time if @xmath60 .
for example , a cosmological model in which the mean densities of galaxies decreases with time can be modeled with @xmath61 and interpreting @xmath12 as the number density of galaxies . in this case ,
equation ( [ newthreek ] ) admits solutions of the form @xmath23 with @xmath62 this equation has the same form as ( [ twof ] ) with the factor @xmath49 replaced by @xmath63 so that the solution is @xmath64 taking the logarithms , followed by the limit @xmath31 , we get the result @xmath65 + \ln a(0 ) .
\label{newlnsum}\ ] ] the convergence of the product in the right hand side ( [ fora ] ) is now more tricky . if @xmath66 becomes close to unity for sufficiently large @xmath21 , then we will pick up a factor @xmath67 in each of the terms . unless this factor is unity , the product in ( [ fora ] )
will either diverge or vanish .
thus we get the condition @xmath68 for convergence thereby making equation ( [ fora ] ) identical to ( [ twof ] ) .
thus , when the normalization condition in ( [ twoint ] ) is relaxed , we again get the same @xmath69dependence as in ( [ twolns ] ) with an extra time ( @xmath21 ) dependence of the form @xmath70 which takes into account the condition ( [ forq ] ) .
this clearly shows that nothing changes qualitatively as far as the spatial dependence is concerned by relaxing ( [ twoint ] ) .
we thus have three possible steady state solutions @xmath71 $ ] _ none _ of which incorporates _ arbitrary _ initial conditions .
obviously if the system is started at any of these solutions , it will stay in it without any evolution .
the question arises as to whether any of them acts as a fixed point for the system evolving from a _ nontrivial _ initial condition or even random poisson initial conditions for which @xmath72 .
while this question is difficult to analyze in the discrete model , it can be answered using a much more detailed description of the system in the continuum limit . in the continuum limit
we need to study the system at two infinitesimally separated moments in time @xmath73 and @xmath74 and obtain a partial differential equation for the evolution of @xmath75 or equivalently for @xmath76 .
we will also need to change the parameters @xmath52 and @xmath7 to @xmath77 and @xmath78 respectively for consistency ; this has the effect of making the _ rate _ of creation and _ rate _ of stretching finite , as it should .
equation ( [ twof ] ) now becomes @xmath79 .
\label{newsix}\ ] ] we have assumed that @xmath43 depends only on @xmath80 making @xmath19 also depend only on @xmath80 . expanding the equation retaining up to linear terms in @xmath81 , and using the result @xmath82 we get the final partial differential equation satisfied by @xmath19 to be @xmath83 the general solution to this equation is straightforward to obtain ; we find that @xmath84\ ] ] where @xmath85 is an arbitrary function of its argument with the condition @xmath86 .
[ this condition incorporates our normalization condition ( [ twoint ] ) .
this condition can be relaxed in exactly the same manner as in the discrete case ; however , as we shall see later , our conclusions do not change . ]
this function - in turn - can be expressed in terms of the initial condition for the problem @xmath87 , which is assumed to be known .
doing this we can write the solution in the form @xmath88 \right ] . \label{dqq}\ ] ] we shall now study the properties of the solution .
let us first consider the simple case in which @xmath43 is given by equation ( [ onew ] ) . in this case
the solution is found to be @xmath89 at any finite @xmath73 , there is a range of @xmath44 values for which the power spectrum ( which is proportional to @xmath90 ) is a power law with the index @xmath91 ) . but note that as @xmath92 the solution decays exponentially at all scales for a wide class of initial conditions ( including poisson distribution with @xmath93 ) .
more generally , if @xmath94 , then the solution in the three ranges go as @xmath95 \cr k^{\beta } \exp \left [ \mu ( \beta - { \lambda \over \mu } ) t \right].\cr}\ ] ] for generic values of the parameters @xmath96 _ all _ the solutions tend to either zero or infinity at late times .
the only exception is if we choose the initial spectrum with @xmath97 .
then , for @xmath98 , we get a pure power law @xmath99 .
this is precisely the solution we found in the discrete case [ see equation ( [ twolns ] ) ] , since in the continuum limit , we can set @xmath100 it must be emphasised that only a very special choice can lead to a nontrivial steady state solution .
even with this special choice we get a solution which has _ no _ time dependence at all .
it is , however , possible to have a power law solution that is valid for a large range of @xmath44 at any finite @xmath73 but with amplitude decreasing exponentially . equation ( [ dqq ] ) can be explicitly integrated for several cases of @xmath43 .
a particularly simple case is the one with @xmath101 in this case the solution is given by @xmath102^{{\lambda \over 2 \mu}}.\ ] ] as @xmath103 , the solution goes to @xmath104 which is again in conformity with the results obtained above , thereby showing that they were not an artifact of the sharp cutoff assumed in equation ( [ onew ] ) . finally , we briefly mention the results in the continuum limit when the normalisation condition ( [ twoint ] ) is relaxed . in this case , the factor @xmath49 in the denominator on the right hand side of equation ( [ newsix ] ) will be replaced by a new constant @xmath50 . this changes equation ( [ partialone ] ) and the solution ( [ dqq ] ) to @xmath105 @xmath106 \right ] .
\label{newdqq}\ ] ] writing the factor inside the integral as @xmath107 it is trivial to see that the solution in this case is the same as the solution with @xmath108 multiplied by the factor @xmath109 $ ] .
that is , @xmath110 .
\label{newfff}\ ] ] this is exactly what one would expect based on our results in the discrete limit in which we found that the solution gets multiplied by a factor in ( [ extrafactor ] ) . in the continuum limit , @xmath111 and @xmath52 gets replaced by @xmath112 and @xmath113 and @xmath21 becomes @xmath114 . in the limit of @xmath115
the extra factor in ( [ extrafactor ] ) becomes @xmath116 \nonumber \\ & = & \exp [ -(\lambda_1 - \lambda ) t ] \end{aligned}\ ] ] which is precisely the extra factor in ( [ newfff ] ) .
thus , even in the continuum limit , our qualitative conclusions do not change when the condition ( [ twoint ] ) is relaxed .
while the discussion above was modeled and motivated by galaxy formation in qssc , the process described here has a much broader range of applicability . the fact that our results were obtained for a model that does not involve gravity explicitly ( for example
, neither newtonian gravitational constant nor the fact that gravitational force varies as a power law is used in the rule " mentioned above ) , suggests that results of the above kind could be quite general .
in fact , there exists several natural phenomena that are described by power law correlations ( see eg .
@xcite ) . more often than not
, such a correlation function seem to arise in a manner that does not depend critically on the details of the underlying dynamical model
. it would be interesting to see whether one can provide a mathematical model with a minimal set of assumptions which can reproduce the power law correlation .
one such minimal set of assumptions can be extracted from the above analysis and we can show that models based on these assumptions will have a generic behaviour . consider a dynamical process in which some physical quantity @xmath117 evolves in time in a manner which depends on its value non locally .
such an evolution can be treated in the discrete version in terms of a rule which allows one to compute @xmath118 in terms of @xmath57 where @xmath21 represents the discretised version of time with , say , @xmath119 and @xmath120 representing a convenient time interval .
for example , the amount of bacteria in a culture , trees in an orchard , buildings in a city or the number of galaxies in some region of the universe could be studied by such prescription . to be more specific
, we shall consider processes of the following kind : we start with a set of points in the d - dimensional space which represents the location of bacteria or galaxies or trees , say .
we now generate a set of new points near each one of the points ( `` non - local self - replication '' ) .
( the bacteria creating new bacteria nearby or trees generating new trees nearby or even cities leading to the formation of new cities nearby seem reasonable . )
let the probability for any given new point to be located at a distance @xmath2 from an old point is @xmath121 .
next , we rescale each of the @xmath122 dimensions by a factor @xmath4 thereby increasing the volume of available space .
( this is useful in the case of growing bacterial culture or a jungle of trees or cities in order to avoid boundary effects which will limit the process . in the cosmological model
mentioned above , this is a natural consequence of the expansion of the universe . )
finally we select a subset of particles in the central region such that the total number of particles remains the same .
this step renormalizes the process back to the original situation so that the process can now be repeated with the new subset of points .
this evolutionary rule can be stated mathematically in the form : @xmath123 \label{geneq}\end{aligned}\ ] ] where @xmath9 are constants with @xmath10 and @xmath11 ; @xmath122 is the dimension of the space in which the vector @xmath13 lives , and @xmath5 is a probability function normalized to unity for integration over all @xmath2 .
this is same as equation ( [ threeq ] ) with @xmath124 replaced by @xmath125 .
we shall also assume that @xmath12 is normalized in such a way that its integral over all @xmath13 is unity .
as discussed above , this can be easily relaxed if required .
it must be stressed that we can consider equation ( [ geneq ] ) as the basic postulate of this analysis rather than any physical model described in the last paragraph ( involving bacteria , trees , cities , galaxies .... ) . in particular : ( i ) we need not restrict to any specific form of @xmath5 or a choice for dimension @xmath122 ; ( ii ) it is also not necessary to identify the vector @xmath13 with the position vector in real space .
the equation ( [ threeq ] ) can also describe quite effectively the power transfer in fourier space when the vector @xmath13 is actually identified with a fourier space vector .
there are phenomena , like fluid turbulence , in which our analysis can be applied by using the power spectrum as the basic variable in @xmath39 space .
the essential postulate will then be that power at nearby wave numbers is generated with a given probability .
this could provide a tool for attacking a wide class of nonlinear phenomena .
two point correlations functions which exhibit power law behaviour are very prevalent in nature .
the analysis here suggests that two ingredients which we have called ( i ) non local self replication and ( ii ) rescaling can lead to such correlation functions fairly generically .
the first of these two ingredients ( non local self replication ) allows similar entities to be created at nearby locations in some space
. this could be as varied as bacterium creating new bacterium nearby by cell division , trees creating trees nearby by seeding or galaxies creating fresh galaxies because of the existence of a creation field .
if we take the space to be the fourier domain , then transfer of power from a given wave number @xmath44 to nearby wave numbers will also constitute such a process .
this variety shows that the ingredient ( i ) is fairly generic and natural .
an immediate consequence of self replication as defined here is the development of correlations .
this is because we have tacitly assumed that the members of the second generation are preferentially ( in the probabilistic sense ) produced near the original parents .
this will naturally lead to the second generation to be correlated with the first , spatially .
if the process takes place in the fourier domain , then the correlation will arise in the space of wave numbers and the consequent relationship in real space will be more complicated ; but even in this case correlation will definitely be established .
( of course , in principle one can also introduce negative correlation in the model by choosing the probability distribution to be an increasing function of distance near the origin . ) in the context of conventional galaxy formation due to gravitational instability , the development of correlation is more indirect and dynamical while the process described here is direct and kinematical . ultimately , the kinematics of the model , encoded in the probability distribution function @xmath126 , need to be connected to an underlying model ( say , the creation field in the case of qssc ) in order to provide the dynamical basis . while this is the basic paradigm in physics ,
it must be stressed that it suffers from two well known difficulties : ( a ) if the physical process is sufficiently complicated ( fluid turbulence , galaxy formation ... ) , it just may not be possible to provide such a dynamical underpinning .
( b ) modeling different processes separately could lead one to miss the existence of a very general description for widely different classes of phenomena .
much of the initial attraction for fractals in the description of natural phenomena originated from its promise to provide such a unifying perspective .
the mathematical formalism developed here should be viewed against such a backdrop .
the second ingredient that we have used ( rescaling ) essentially serves to renormalise the scales and is , in fact , very similar to ideas used in the theory of renormalization group .
once self replication takes place , the system has become denser on the whole and in order to concentrate on the intrinsic correlation , it is necessary to renormalise the system back to the original state . in all the examples which we have it is always possible to link this rescaling with a tangible physical process .
the combination of these two ingredients very nicely and naturally leads to a system with higher level of correlation at each time step .
in fact , at least in the case of galaxy formation , the combined effect of these two phenomena is very similar to actual gravitational attraction between the particles .
the key result of the analysis is that processes with these two ingredients are inherently unstable in the sense that the correlation function either grows without bound or decays to zero ( with the system becoming more and more dilute ) as @xmath127 .
once again such an instability is reminiscent of similar phenomena seen in self gravitating systems though we have not used any gravitational dynamics .
there is one special initial condition which leads to a static solution but as we discussed in the text , this is rather too special .
the interest in these two systems lies in the intermediate time scale during which it could exhibit a power law correlation function very generically .
once again the situation is similar to galaxy formation due to gravitational instability in which the observed power law correlation function will exist _ only _ during a limited temporal and spatial window in the numerical simulations . even in the case of fractals , it is generally known that one has to introduce cut - off in spatial and temporal scales in order to maintain power law correlation functions . in conclusion , it is interesting how a fairly simple mathematical model could lead to an approximate description of a wide class of phenomena
. only further investigations for specific contexts will determine whether such a unified treatment is of some value or whether these ideas are destined to remain as mathematical curiosities .
i thank j.v .
narlikar for comments on an earlier version of the manuscript .
i appreciate the detailed and constructive comments from an anonymous referee which significantly improved the presentation of the ideas .
padmanabhan t. , 1996 , mnras , 278 , l29 [ astro - ph-9508124 ] .
nayeri , a. engineer , s. narlikar , j. v. hoyle , f. , 1999 , ap j , 525 , 10 .
mandelbrot , b. b. , 1983 , fractal geometry of nature , ( w.h .
freeman , new york ) . | a large class of evolutionary processes can be modeled by a rule that involves self - replication of some physical quantity with a non local rescaling .
we show that a class of such models are exactly solvable in the discrete as well as continuum limit and can represent several physical situations as varied from the formation of galaxies in some cosmological models to growth of bacterial cultures .
this class of models , in general , has no steady state solution and evolve unstably as @xmath0 for generic initial conditions
. they can however exhibit ( unstable ) power law correlation function in the continuum limit , for an intermediate range of times and length scales . |
the standard jaynes - cummings model ( jcm ) is usually referred to as the simplest fully quantized light - matter interaction scheme involving just one atom and one mode of the radiation field . since its introduction in 1963 @xcite , the model has attracted a great deal of theoretical and experimental interest in the fields of laser physics and quantum optics @xcite , and this trend is due in part to its apparent simplicity and , more importantly , to its striking predictions about the dynamical properties of the subsystems involved with which the quantum optics community is now familiar , as for example the well - known phenomenon of _ collpases _ and _ revivals _ for the atomic excitation understood in terms of the grainy statistics of the radiation field @xcite .
+ the jcm has become the source of inspiration for a wide variety of generalizations dealing with more general and/or realistic circumstances .
the majority of them focused primarily upon multi - photon transitions and/or multimode fields @xcite ; engineering nonlinear atom - field couplings , like the buck - sukumar and kochetov models @xcite ( including some very recent improvements and extensions of the former with or without applying the rotating wave approximation @xcite ) , one - atom jc models involving two - photon interaction with intensity - dependence both in the atom - field coupling and the detuning @xcite , or adding nonlinear kerr - like media @xcite ; interacting or noninteracting two two - level atoms @xcite ; or even more complex systems involving a large group of @xmath0 two - level ( or multi - level ) atoms in the same cavity , such as the so - called tavis - cummings model ( tcm ) @xcite , to mention some examples . and still more recently , renewed attention has been paid to quantum decoherence and entanglement properties of light - matter interaction models la jaynes - cummings whose central system is composed of two or three two - level atoms ( also called within the jargon of the quantum information framework as two- or three - qubit systems ) resonantly coupled with a cavity field prepared in a single number state and also with each other through dipole - dipole and ising - like interactions @xcite ( incidentally , spin - spin interactions such as these or similar have also been the theme of current interest in a manifold of areas such as optical lattices @xcite ; systems with trapped ions @xcite and microcavities @xcite ; and , even in the context of nearly localized and dipolarly coupled two identical molecules @xcite , where , under certain conditions , from an algebraic - structure point of view , intramolecular coupling models la jaynes - cummings emerge ) ; moreover , along this line an interesting application based on the resonant two - atom jcm has been proposed with the aim of implementing novel protocols for unambiguous bell state discrimination for two qubits @xcite .
+ this work is in keeping with the aforementioned spirit of putting forward another extension of the jaynes - cummings model , that is , a nonlinear version of it for the description of a qubit system composed of two identical two - level atoms that interact with each other via dipole - dipole and ising - like interaction and with a one - mode cavity field being in a coherent superposition of number states .
the nonlinear character of the model is construed in a twofold sense : i ) the interaction between the atomic system and the field is considered to be nonlinear in a way such that it hinges on the number of photons of the latter , and ii ) the said atomic system can be allowed to be embedded in a kind of nonlinear medium inside the cavity . in considering a buck - sukumar nonlinear coupling @xcite between the atoms and the field and a kerr - like medium within the cavity as particular cases , a full algebraic solution of the problem is provided and the consequences of incorporating such nonlinearities are illustrated by exploring , in the resonant quantum dynamics , some quantities of current physical interest , such as population inversion , purity , and concurrence , as far as the atomic system is concerned , and the time behavior of the field itself is also investigated in terms of its entropy and its image on phase space . it is worth commenting that the set of results reported here , regarding the aforesaid nonlinear coupling scheme , may also be of some relevance in the light of novel experimental and theoretical research on optical simulation of the jaynes - cummings and rabi models in arrays of coupled photonic waveguides @xcite , as well as in current designs of architectures intended for quantum computation and communication based on cavity quantum electrodynamics ( cqed ) studies for solid - state superconducting electrical circuits @xcite .
furthermore , we should stress the fact that the present treatment is not only valid for interactions between quantized fields and atoms , but also in the context of trapped ions interacting with lasers ( classical ) fields where a number of phenomena known from nonlinear optics , including kerr - like nonlinearities , can be easily produced ( see , for instance , refs .
so , from our point of view , the foregoing contributions may open up the possibility of realizing hamiltonian models such as the one we are considering in physically different but algebraically equivalent contexts .
+ the content of the paper is the following . in section 2 the hamiltonian model and its general solution for two ising- and dipole - dipole interacting and identical two - level atoms
are introduced .
section 3 , which in turn is divided into three subsections , is thoroughly devoted to the discussion of some properties of the atomic system in terms of its population , purity and concurrence dynamics ; all these features are explored by taking into consideration each nonlinear contribution of the model . in section 4 ,
the effect of such nonlinearities on the evolution of the cavity field is explored based upon its entropy and its phase space picture via the q - function . and
finally , in section 5 some conclusions are given .
let the hamiltonian model of the system under study be encoded as follows : @xmath1 where the constituent hamiltonians are explicitly given , in the schrdinger picture , and under the rotating - wave approximation , by @xmath2 here , @xmath3 and @xmath4 are the energy operators for the field and atoms , respectively , the coupling between the atomic system and the radiation field is described by @xmath5 , and @xmath6 is the contribution to the total system of the atom - atom interaction mentioned above .
the field mode frequency is @xmath7 , @xmath8 is the atomic transition frequency , @xmath9 is the coupling constant ( which is taken to be the same for both atoms ) , and @xmath10 and @xmath11 are , respectively , the ising and dipole - dipole parameters . as usual ,
@xmath12 ( @xmath13 ) is the photon annihilation ( creation ) operator satisfying @xmath14=1 $ ] , and , on the other side , @xmath15 ( @xmath16 ) , @xmath17 ( @xmath18 ) are the standard atomic two - level transition operators of the respective atoms .
finally , @xmath19 and @xmath20 , with @xmath21 , are photon - number - dependent functions that represent the aforesaid nonlinear character of the whole system . at this point , we note in passing that although we shall be working with parameters such that the rotating - wave approximation is considered to be valid , we also have to be careful with the field intensities reflected upon the functions @xmath20 and @xmath19 , always taking into account that the conditions @xmath22 and @xmath23 are fulfilled and that the effective detuning @xmath24 still allows us to work within the two - level atom approximation . the origin of and the role played by such an intensity - dependent detuning can readily be seen if we reframed our system under the transformation @xmath25 to get an atom - field coupling structure of the form @xmath26 , with the fitting identification @xmath27 of the detuning we are referring to .
the intensity - dependent part of the latter is construed as the frequency separation between adjacent energy levels of a nonlinear field ( i.e. , a nonlinear oscillator , viewed as a whole ) whose spectrum is clearly nonequidistant by virtue of @xmath19 in the definition of the field hamiltonian @xmath3 . + here
, we find it convenient to choose the framework generated by the unitary operator @xmath28 $ ] so that one can recast hamiltonian ( [ eq : totalham ] ) , on inserting it into the transformation @xmath29 as follows @xmath30 where we have labeled @xmath31 and @xmath32 .
it is well - known that the dynamics of a system such as this involving a single - photon process where the number of excitations is a conserved quantity ( as can be verified since the excitation number operator @xmath33 commutes with hamiltonian ( [ eq : totalhamtrans ] ) ) the set of states defined by @xmath34 may be employed as a basis to diagonalize the hamiltonian under study ( see , for instance , refs .
however , one can see that it is also possible to employ the following fitting basis : @xmath35 and @xmath36 turns out to be convenient in the sense that it will allows us to simplify subsequent calculations ( such a symmetric state , needless to say , is useful only in the particular case when the two atoms are identical ) .
so , in this basis the representation of @xmath37 is a @xmath38 matrix with elements @xmath39 for a given number of photons @xmath40 . in units
where @xmath41 we get : @xmath42 where , for the sake of simplicity , we have utilized the shorthand notation @xmath43 , @xmath44 , with @xmath45 .
+ for all the above initial considerations , we now proceed to solve the eigenvalue equation @xmath46 for @xmath47 . tackling
this problem entails solving a cubic polynomial equation whose roots are well - known .
so , it follows straightforwardly from cardano s formulae that the sought eigenvalues @xmath48 are given by @xmath49 with @xmath50 such eigenvalues are all real provided that @xmath51 , with @xmath52 , @xmath53 and @xmath54 being , respectively , @xmath55 where we have set the photon - number dependent functions : @xmath56
so , we let the corresponding set of normalized eigenvectors be written as @xmath57 for @xmath47 , and where the normalization factor @xmath58 reads @xmath59 in order to further simplify subsequent algebraic manipulations we let ( [ eq : vector ] ) be recast as @xmath60 with the following correspondences @xmath61 from which one can also deduce that @xmath62 this set of results will be our starting point for examining some properties of physical interest concerning the evolution of either the atomic system or the cavity field , such as the population inversion , entanglement dynamics based on the purity and concurrence features , and the evolution of the field on phase space with the help of the q - function representation .
let us first examine the effect of dipole - dipole and ising - like interactions on the dynamics of the atomic level occupation .
the time evolution of this physical feature can be assessed by considering the inversion operator @xcite @xmath63 together with the use of the wave function @xmath64 where @xmath65 is the initial state of the whole system . for the time being , let the initial state be in a way such that the atoms are both in their corresponding excited state and the cavity field in a coherent superposition of number states , i.e. , @xmath66 which , by utilizing ( [ eq : basis2 ] ) , can be rewritten as @xmath67 so that @xmath68 where we have set the time - dependent coefficient @xmath69 by taking the expectation value of ( [ eq : opdz ] ) with the help of ( [ eq : wavefunc ] ) , and after making some rearrangements , we arrive at the desired result for the atomic inversion @xmath70 in which @xmath71 represents the initial photon distribution function of the field , @xmath72 with @xmath73 and @xmath74 being given by ( [ eq : q ] ) and ( [ eq : theta ] ) , respectively . and , following a similar notation as in ref .
@xcite , the weighting amplitudes @xmath75 s take the form @xmath76 where , in turn , the @xmath77 and @xmath78 are given by ( [ eq : cj1 ] ) and ( [ eq : cj3 ] ) .
+ from eq .
( [ eq : inversion ] ) one can readily see that , as opposed to the well - known one - atom model , the dynamics of atomic inversion consists , save for a constant term , essentially of a superposition of three different oscillatory components , each giving rise to three different patterns of collapses and revivals and characterized by its own rabi frequency ( see eqs .
( [ eq : rabi1])-([eq : rabi2 ] ) ) .
it is found that this set of frequencies satisfies , in general , the following relationships : @xmath79 @xmath80 where the vertical bars @xmath81 denote the absolute value of the quantity involved and @xmath82 + \nonumber \\ \fl & & + 2\omega_{0}(j-\kappa)(2f_{n,1}-f_{n,0}-f_{n,2})+3\omega_{0 } \delta ( f_{n,2}-f_{n,0 } ) \nonumber \\ \fl & & -4(j-\kappa)^{2}-3\delta^{2}-6g^{2}\delta_{n}^{+ } \bigg \}.\end{aligned}\ ] ] a better picture of the evolution of this physical feature will be shown by the graphical representation of two particular cases : firstly , we will consider a situation in which the atomic system is supposed to be immersed in a nonlinear kerr - like medium within the cavity , so that the number operator function @xmath83 in eq .
( [ eq : hamfield ] ) is chosen to have the form @xmath84 , leading to the hamiltonian of the field @xmath85 ; here , @xmath86 is construed as an anharmonicity parameter associated with the dispersive part of the third - order nonlinearity of the medium within which the radiation field also evolves @xcite .
[ this is , incidentally , an approximate and justified model widely used in nonlinear optics @xcite whenever the medium s time response is sufficiently small so that the medium itself may be regarded as being able to follow the field in an adiabatic manner @xcite ; an agreeable and thorough treatment of the subject had already been undertaken , for instance , by buzek and jex @xcite in their work on the jaynes - cummings model with only one atom when the cavity is supposed to be filled with a kerr - like medium ] .
secondly , the influence of the interplay between the two atoms ( @xmath87 ) upon the population dynamics is explored by taking into consideration a standard cavity , i.e. , when @xmath88 ( @xmath89 ) . unless otherwise specified , in subsequent calculations the initial state of the field is considered to be a coherent state with a poissonian statistiscal distribution @xmath90 with @xmath91 photons .
merging the nonlinearity of the medium within the cavity with both the dipole - dipole and ising - like atomic interactions leads to the following analytic expressions for cardano s formulae @xmath73 and @xmath74 ( eqs .
( [ eq : q ] ) and ( [ eq : theta ] ) ) in terms of the parameters @xmath86 , @xmath11 , and @xmath10 : @xmath93 + 54 g^{2}\chi ( n+1)\delta_{n}^{-}}{\left ( \left(\chi-2(\kappa - j)\right)^{2}+12 \chi^{2}(n+1)^{2}+6g^{2}\delta_{n}^{+ } \right)^{3/2 } } \right \ } , \nonumber\end{aligned}\ ] ] where the number - dependent function @xmath94 describing the nonlinear dipole coupling between the field and the atoms remains implicit into the frequency shifts @xmath95 and the cavity frequency is assumed to match the atomic transition frequency ( @xmath96 ) .
it is worth highlighting that this set of equations can also be viewed as dependent on the difference @xmath97 for a given @xmath86 , instead of focusing separately on the values of @xmath11 and @xmath10 .
such a dependency is of course reflected upon the rabi frequencies ( [ eq : rabi1])-([eq : rabi3 ] ) and the set of weighting amplitudes ( [ eq : amplitudes ] ) by direct substitution of @xmath73 and @xmath74 given above .
it is clear then that the dipole - dipole and ising interactions cancel each other when the equality @xmath98 holds , as though there was a kind of trade - off between both atomic interactions . + the evolution of atomic excitation exhibits its characteristic collpase - and - revival behavior ( explained in terms of the granular statistics of the field @xcite ) but in a not - so - conventional fashion , depending on whether the nonlinear character of the medium is featured or whether the interplay of the atoms , together with the nonlinear coupling between them and the radiation field , is taken into account in our description .
for instance , one can see from the sequence of graphs in fig .
[ fig : atinversion1 ] ( a ) that in the case of a linear coupling , for which @xmath99 , and under the condition @xmath100 , the appearance of revivals in the dynamics of atomic excitation tends to be more frequent insofar as the value of @xmath86 increases , and the profile of such revivals becomes sharpened and squeezed in the meanwhile . following this sequence ,
the results also reveal the fact that the offset created by the constant term in the atomic excitation turns out to be highly dependent on @xmath86 ; the larger its value , the more significant its contribution to the deviation from the abscissa where @xmath101 . on the other hand , adding a buck - sukumar - like function @xmath102 to the coupling between the atoms and the field fuels significantly the foregoing behavior as far as the profile and the number of revivals we observe , although in this case their frequency of appearance remains almost unchanged provided we keep ourselves confined to these restricted values of @xmath86 ( see fig . [
fig : atinversion1 ] ( b ) ) . comparatively , the offset seems to be less sensitive to the combined effect of both nonlinear contributions . + for an initially coherent state with @xmath103 photons .
the parameters are @xmath96 , @xmath104 , @xmath105 , with @xmath84 , @xmath99 ( left panel ) and @xmath102 ( right panel ) .
the following cases are considered : @xmath106 , @xmath107 , @xmath108 , and @xmath109 , corresponding to red , blue , green and black curves , respectively.,title="fig:",width=283,height=264 ] for an initially coherent state with @xmath103 photons .
the parameters are @xmath96 , @xmath104 , @xmath105 , with @xmath84 , @xmath99 ( left panel ) and @xmath102 ( right panel ) .
the following cases are considered : @xmath106 , @xmath107 , @xmath108 , and @xmath109 , corresponding to red , blue , green and black curves , respectively.,title="fig:",width=283,height=264 ] some structure in the overall atomic population profile can be found if the anharmonic regime of the cavity field , in combination with the interatomic interplay and the current intensity dependent atom - field coupling , is such that the condition @xmath110 holds .
let us consider a situation in which the relationship between the aforesaid anharmonicity and the atomic interplay is such that @xmath111 .
this particular scenario allows us to simplify further our algebraic results and , more importantly , quantify in some detail the dynamics of the atomic population when the above - mentioned influences are compounded .
so , under the additional condition @xmath112 ( i.e. , in the limit of moderately high field intensities ) , one gets , in terms of what we have labeled , for the time being , as the scaled anharmonicity parameter @xmath113 ( with @xmath9 being the coupling constant ) , the following effective approximations for the weighting amplitudes ( [ eq : amplitudes ] ) : @xmath114 which are taken to be common to all oscillatory components in ( [ eq : inversion ] ) around @xmath115 .
likewise , for the ones associated with the offset s contribution we obtain @xmath116 we now proceed to provide an approximate version of the rabi frequencies @xmath117 and @xmath118 , regarded as functions of the independent variables @xmath119 and @xmath40 , by expressing them in terms of their taylor s series expansion about the point @xmath120 ; this is so because , as functions of @xmath40 , such frequencies are supposed to be strongly weighted around @xmath115 by virtue of the poissonian statistics of the initial coherent field .
that is , @xmath121+o(x^{3},n^{3 } ) .
\nonumber \end{aligned}\ ] ] based on ( [ eq : rabi2 ] ) and ( [ eq : rabi3 ] ) , together with the set of equations ( [ eq : cardanosqt ] ) , the corresponding expansion coefficients , up to second order in @xmath119 and @xmath40 , take the form @xmath122 here , aside from the homogeneous terms @xmath123 , @xmath124 , and @xmath125 , each derivative of @xmath126 with respect to @xmath40 is known to define different time scales @xcite : the first order derivatives are responsible for the revival behavior whose periodicity is easily determined to be @xmath127 , whereas , at least in the regime we are working on , the second and higher order ones give rise to dephasing effects that become significant over much longer time scales .
so , after substituting ( [ eq : domei1])-([eq : domef1 ] ) into ( [ eq : series1 ] ) , we arrive , up to first order in @xmath40 , at the sought approximations @xmath128 where @xmath129 is an average - photon - number dependent phase shift ; this behaves approximately as a linearly increasing function of @xmath115 for @xmath102 , hence even second order contributions of @xmath119 can not be ignored for sufficiently high field intensities . on the basis of this set of approximations ,
we finally work out that the atomic inversion takes the form @xmath130
\nonumber \\ & & + \lambda_{23}^{(x ) } \cos \left [ 3(1-x)\tau + \varphi_{\langle n \rangle}x^{2}\tau+\langle n \rangle \sin(2\tau ) \right ] \bigg \ } , \label{eq : invapproxchi}\end{aligned}\ ] ] where @xmath131 . in fig .
[ fig : atinversion1chi ] we show , within the scaled interval @xmath132 , the outcome of this approximation ( frame ( a ) ) that is in agreement with the exact one coming directly from eq .
( [ eq : inversion ] ) ( frame ( b ) ) whenever @xmath133 holds ; needless to say , this is so since the removal of second order contributions of @xmath40 in the taylor series demands such a restriction .
the evident aspect of the atomic evolution is that it exhibits a kind of beating behavior enclosing , in turn , a series of revivals of shorter duration .
( [ eq : invapproxchi ] ) , the periodicity of the beating modulation , as well as the collapse and revival times corresponding to the confined revivals , can be approximately quantified . in this regard ,
we get @xmath134 , with @xmath135 being an integer , for the modulation frequency , whereas @xmath136 and @xmath137 for the inner revival and collapse times , respectively .
this set of calculations were performed by considering a scaled anharmonicity parameter @xmath138 and a coherent field with @xmath139 photons . + it is worth highlighting the fact that even a very slight and/or inconspicuous deviation from the standard cavity produced by the kerr - type nonlinearity , necessarily combined with an intensity dependent atom - field coupling , can produce a somewhat well - structured population dynamics . on the the contrary
, if the the nonlinear character of the atom - field coupling is switch off , i.e. , @xmath99 , the overall behavior looks like that in fig .
[ fig : atinversion1 ] ( a ) and seems to be , in turn , qualitatively similar to the one - atom case reported in the work of gra and jedrzejek @xcite , a regime under which the structures described above do not take place . and , as mentioned at the outset of this work
, our model not only goes beyond the one - atom case to consider the one of two interacting atoms but also would give us the possibility of exploring any other regime and/or physical scenario in which more complex atom - field couplings may be engineered @xcite . calculated from eqs .
( [ eq : invapproxchi ] ) ( red line ) and ( [ eq : inversion ] ) ( blue line ) in the buck - sukumar regime ( @xmath102 ) and interacting atoms within a kerr - like cavity such that @xmath140 , with the field being initially in a coherent state having an average photon number @xmath139 . , title="fig:",width=283,height=170 ] calculated from eqs .
( [ eq : invapproxchi ] ) ( red line ) and ( [ eq : inversion ] ) ( blue line ) in the buck - sukumar regime ( @xmath102 ) and interacting atoms within a kerr - like cavity such that @xmath140 , with the field being initially in a coherent state having an average photon number @xmath139 . , title="fig:",width=283,height=170 ] as to the standard cavity field , for which @xmath89 , cardano s formulae become @xmath142
so , using these expressions enables us to deploy eq .
( [ eq : inversion ] ) in a way such that it can be viewed as a function of the difference @xmath97 , as we have mentioned before .
the dynamics of the atomic occupation is displayed in fig .
[ fig : atinversion2 ] for @xmath143 , @xmath107 , @xmath108 and @xmath109 . in this regime , concerning the case in which @xmath99 ( see fig .
[ fig : atinversion2 ] ( a ) ) , the atomic system exhibits a somewhat similar collapse - and - revival behavior to that in the well - known standard two - atom jaynes - cummings model @xcite , save for the fact that it is observed a faster blurring of the train of revivals with the increase of the difference @xmath97 .
however , we see a completely different pattern when both the dipole - dipole - and - ising interactions and the buck - sukumar coupling ( @xmath102 ) are compounded ( see fig . [
fig : atinversion2 ] ( b ) ) .
it is interesting to note that both contributions , working together , foster the formation of well - structured profiles of beats composed of a number of inner revivals , as seen , for instance , more remarkably from the red and blue graphs in fig .
[ fig : atinversion2 ] ( b ) corresponding to the cases when @xmath143 and @xmath107 , respectively ; one sees that this is not an exclusive behavior of the kerr effect observed previously .
we also point out that neither the dipole - dipole - and - ising interactions nor its combination with the nonlinear atom - field coupling play a proponderant role in the atomic excitation s offset .
+ motivated by the above results involving both the nonlinear atom - field coupling ( @xmath102 ) and the interatomic interplay within the regime @xmath144 ( which is , in fact , slightly less restrictive than that of the kerr - like cavity ) where the evolution of atomic excitation reveals what seems to be a regular and well - structured profile , we undertake the task of looking for a suitable approximation in order to gain some insight into such a behavior .
it turns out that , for @xmath112 , the weighting amplitudes given by ( [ eq : amplitudes ] ) are found to be well approximated as follows : @xmath145 and @xmath146 . on the basis of eqs .
( [ eq : qnkj ] ) and ( [ eq : tetankj ] ) , let us now make a taylor expansion of the surviving rabi frequencies ( [ eq : rabi2 ] ) and ( [ eq : rabi3 ] ) in @xmath147 and @xmath40 by employing eq .
( [ eq : series1 ] ) to get explicitly the corresponding expansion coefficients up to second order in @xmath119 and @xmath40 : @xmath148 here , we have set @xmath149 and all derivatives are evaluated at the point @xmath150 ; additionally , we have applied the fitting approximation @xmath151 . at this point , as opposed to the kerr - cavity case , given the inverse proportionality of ( [ eq : domef ] ) as a function of @xmath152 it makes sense to retain only contributions up to first order in @xmath119 in the expansion , thereby getting @xmath153 where we have also found it convenient to cut the series up to first order terms in @xmath40 , as long as we restrict ourselves to proper time scales . on substituting all the foregoing approximate expressions into eq .
( [ eq : inversion ] ) , we obtain the sought result @xmath154 where we have set @xmath155 .
this expression is , in turn , summed exactly to arrive at the following closed form @xmath156 .
\label{eq : invapproxkj2}\ ] ] based on this approximate result , the evolution of atomic excitation can be understood in terms of the product of what we can identify as an envelope factor , @xmath157 , and a more rapidly oscillating cosine function .
the essential feature of the former is twofold as depicted in fig .
[ fig : atinversion3 ] ( a ) .
i.e. , on the one hand , it transpires that the exponential part of it takes part in enveloping the inner revivals generated by the term of higher frequency @xmath158 $ ] ; incidentally , the argument of the latter , in contrast to the one - atom model , contains an additional time dependent term , @xmath159 , that is in fact responsible for the alternating phase inversion of successive revivals at the first stages of atomic evolution , as seen in fig .
[ fig : atinversion3 ] ( b ) where , for the sake of comparison , we have also depicted the exact result obtained by means of eq .
( [ eq : inversion ] ) for the particular case of @xmath160 in the scaled interval @xmath161 . on the other hand
, it follows that the dipole - dipole and/or ising interactions constrain the overall modulation amplitude ( outer , red , dashed line in fig .
[ fig : atinversion3 ] ( a ) ) via the enveloping term of lower frequency , @xmath162 , thereby giving rise to the above - mentioned characteristic train of beats ; their periodicity is easily determined to be held at times @xmath163 , with @xmath135 being an integer .
we have plotted in fig .
[ fig : atinversion3 ] ( c ) the exact result ( blue , continuous line ) in the same interval as that of the approximate version of it in fig .
[ fig : atinversion3 ] ( a ) , i.e. , @xmath164 , where the qualitative behavior of the former is well reproduced by the latter even at such time scales .
finally , both the exact ( blue , continuous line ) and approximate ( red , dashed line ) results are displayed together in fig .
[ fig : atinversion3 ] ( d ) in a much more faraway time interval , where it is clear that the outcome of eq .
( [ eq : invapproxkj2 ] ) ( red , dashed line ) fails at these far longer times .
this noticeable discrepancy between the approximate and the exact results is due in part to the asymptotic approximations involving the weighting amplitudes and the number - dependent frequency shifts @xmath165 given above and primarily to the removal of the second order contributions of @xmath40 in the taylor expansion of the rabi frequencies .
based on ( [ eq : series1 ] ) and ( [ eq : tscalen ] ) , as long as we restrict ourselves to time scales such that @xmath166 , with @xmath155 , eq .
( [ eq : invapproxkj2 ] ) gives us an illustrative and quite acceptable first approximation and becomes more and more accurate for longer periods of time as the coherent - state average photon number @xmath167 increases . and , in exactly the same way as in the kerr cavity case , one can extract from it some useful information concerning the inner collapse and revival times , namely , @xmath168 and @xmath169 , respectively , and , it is of some importance to remark that both of them turn out to halve their corresponding value in the one - atom case @xcite .
this specific set of calculations was carried out by considering the initial state of the field as being in a coherent state with @xmath170 photons . + for an initially coherent state with @xmath103 photons .
the parameters are @xmath96 and @xmath89 , with @xmath88 , @xmath99 ( left column ) and @xmath102 ( right column ) .
the following cases are considered : @xmath171 , @xmath107 , @xmath108 , and @xmath109 , corresponding to red , blue , green , and black curves , repsectively.,title="fig:",width=283,height=264 ] for an initially coherent state with @xmath103 photons .
the parameters are @xmath96 and @xmath89 , with @xmath88 , @xmath99 ( left column ) and @xmath102 ( right column ) .
the following cases are considered : @xmath171 , @xmath107 , @xmath108 , and @xmath109 , corresponding to red , blue , green , and black curves , repsectively.,title="fig:",width=283,height=264 ] calculated from eqs .
( [ eq : inversion ] ) ( blue , continuous line ) and ( [ eq : invapproxkj2 ] ) ( red , dashed line ) in the buck - sukumar regime ( @xmath102 ) and interacting atoms ( @xmath143 ) with the field being initially in a coherent state with average photon number @xmath139 .
frames ( a ) and ( c ) show the evolution of atomic excitation within the time interval @xmath172 . frames ( b ) and ( d ) encompass two widely separated time intervals , @xmath161 and @xmath173 , respectively , so as to emphasize the limitation of the approximation ( [ eq : invapproxkj2]).,title="fig:",width=283,height=151 ] calculated from eqs .
( [ eq : inversion ] ) ( blue , continuous line ) and ( [ eq : invapproxkj2 ] ) ( red , dashed line ) in the buck - sukumar regime ( @xmath102 ) and interacting atoms ( @xmath143 ) with the field being initially in a coherent state with average photon number @xmath139 .
frames ( a ) and ( c ) show the evolution of atomic excitation within the time interval @xmath172
. frames ( b ) and ( d ) encompass two widely separated time intervals , @xmath161 and @xmath173 , respectively , so as to emphasize the limitation of the approximation ( [ eq : invapproxkj2]).,title="fig:",width=283,height=151 ] calculated from eqs .
( [ eq : inversion ] ) ( blue , continuous line ) and ( [ eq : invapproxkj2 ] ) ( red , dashed line ) in the buck - sukumar regime ( @xmath102 ) and interacting atoms ( @xmath143 ) with the field being initially in a coherent state with average photon number @xmath139 .
frames ( a ) and ( c ) show the evolution of atomic excitation within the time interval @xmath172
. frames ( b ) and ( d ) encompass two widely separated time intervals , @xmath161 and @xmath173 , respectively , so as to emphasize the limitation of the approximation ( [ eq : invapproxkj2]).,title="fig:",width=283,height=151 ] calculated from eqs .
( [ eq : inversion ] ) ( blue , continuous line ) and ( [ eq : invapproxkj2 ] ) ( red , dashed line ) in the buck - sukumar regime ( @xmath102 ) and interacting atoms ( @xmath143 ) with the field being initially in a coherent state with average photon number @xmath139 .
frames ( a ) and ( c ) show the evolution of atomic excitation within the time interval @xmath172 .
frames ( b ) and ( d ) encompass two widely separated time intervals , @xmath161 and @xmath173 , respectively , so as to emphasize the limitation of the approximation ( [ eq : invapproxkj2]).,title="fig:",width=283,height=151 ] from the point of view of quantum information , another feature that deserves to be examined is the entanglement dynamics of the two interacting atoms . to do this ,
let us now suppose that our atomic system is initially prepared in the symmetric entangled state given , for a definite number of photons @xmath40 , by eq .
( [ eq : state2 ] ) . in doing so , and following the same procedure as that outlined above for the atomic excitation
, one finds that the state of the composite system evolves as follows : @xmath174 where , again , the state of the cavity field is supposed to be in a general superposition of number states , each of them weighted by the set of amplitudes @xmath175 , and the time - dependent coefficient takes now the form @xmath176 thus , starting from the corresponding density operator of the atom - field system , namely , @xmath177 we first proceed to calculate the purity of the atomic system alone by means of the standard definition of this property @xmath178 where @xmath179 represents the reduced density operator of the atoms computed via @xmath180 with @xmath181 denoting the trace over the field variables .
hence , one obtains for the reduced density operator @xmath182 where the set of atomic states @xmath183 comes from having made use of a shorthand notation for the basis ( [ eq : state1])-([eq : state3 ] ) , namely , we have labeled @xmath184 with the following correspondences : @xmath185 , @xmath186 , and @xmath187 .
then , on substituting the last expression into ( [ eq : purityatom ] ) , we get @xmath188 in this context , the expression for the purity given above can be understood as a tool giving an account of the degree of entanglement between the atomic system and the cavity field , with the latter being viewed as the environment , and of particular interest to us will be , again , a radiation field in a coherent superposition of number states .
+ the evolution of the purity of the atomic system for a field initially prepared in a coherent state with @xmath189 , is depicted in fig .
[ fig : purity - cs ] encompassing the nonlinear regimes considered so far .
more precisely , the figure aims to stress the role of the nonlinear atom - field coupling ( i.e. , the buck - sukumar regime in which the coupling function @xmath102 ) as well as the inherent nonlinearity of the kerr - like cavity in the evolution of the aforementioned quantity .
figures [ fig : purity - cs ] ( a ) and ( b ) show , respectively , the corresponding outcomes when either a linear or a nonlinear atom - field coupling is taken into consideration .
frame ( a ) indicates , for instance , that within the linear regime ( @xmath99 ) of the standard cavity the central system of interacting atoms ( @xmath190 ; gray curve ) tends to be more robust against decoherence at the first stages of its evolution in opposition to the kerr - like cavity case ( @xmath191 ; black curve ) with non - interacting atoms ( @xmath105 ) . even though both cases display a kind of spiky behavior as time elapses , in the kerr - like cavity the appearance of small , distinguishable and almost regular spikes revelas a very weak attempt to restore the initial coherence of the atomic system .
this behavior seems to be reinforced and much more frequent when the intensity dependent coupling is featured ( @xmath102 ) , as seen in frame [ fig : purity - cs ] ( b ) on a shorter time scale .
i.e. , we see that the atom - atom interaction in conjunction with the buck - sukumar coupling ( gray curve ) strengthens the possibility of recovering almost completely the initial phase information of the two atoms at certain and more regular time instants .
when both the kerr - like cavity and the nonlinear coupling are working together ( frame ( b ) ; black curve ) , the atomic system becomes most of the time less robust against decoherence , but it exhibits , in a haphazard fashion , more frequent and pronounced spikes than those appearing in the case of linear coupling ( see frame ( a ) ; black curve ) . regarding the dipole - dipole and ising - like atomic interactions , it has been verified that their influence upon the purity dynamics of the atoms themselves do not give rise to significant changes in the general behavior of it at least within the range @xmath192 , as shown in fig .
[ fig : purity - cs2 ] ( a ) for three different time instants , namely , @xmath193 , @xmath194 and @xmath195 .
each of these correspond respectively , and approximately , to the second , third and fifth maximum in the evolution of purity shown in fig .
[ fig : purity - cs ] ( b ) for the buck - sukumar - like coupling case ( gray curve ) .
so , as a function of the effective atomic parameter @xmath147 , we can see that the purity decreases gradually when we extend the range of values of @xmath97 ; secondarily , in the regime when @xmath196 , the profile of the atomic purity as a function of time was also found to undergo an inconspicuous phase shift with respect to the case when @xmath105 just at the first stages of the evolution .
a somewhat different behavior is observed in the case of a kerr - like cavity illustrated in fig .
[ fig : purity - cs2 ] ( b ) for the same time instants . from the figure
we see that what started off being an almost pure atomic state ended up being a state that loses its initial coherence with the increase of @xmath197 , and the decay of this atomic coherence turns out to be far more pronounced than that of the previous case .
by contrast , far beyond the range of values of the anharmonicity parameter considered so far , say , when @xmath198 , the atomic purity seems to display a raising tendency as the aforesaid parameter increases .
we surmise that this last conduct may be closely linked to the atomic population trapping effect @xcite induced by the presence of the nonlinear medium through a kind of detuning effect at such high values of @xmath197 . + ) for the field being initially in a coherent state with an average photon number @xmath103 .
the parameters are @xmath190 ( gray line ) and @xmath199 ( black line ) , with @xmath99 ( frame ( a ) ) and @xmath102 ( frame ( b ) ) .
, title="fig:",width=287,height=188 ] ) for the field being initially in a coherent state with an average photon number @xmath103 .
the parameters are @xmath190 ( gray line ) and @xmath199 ( black line ) , with @xmath99 ( frame ( a ) ) and @xmath102 ( frame ( b ) ) .
, title="fig:",width=287,height=188 ] ) plotted , separately , as a function of @xmath147 ( frame ( a ) ) and @xmath197 ( frame ( b ) ) for the time instants @xmath193 ( green ) , @xmath194 ( red ) , and @xmath195 ( blue ) .
the field is initially in a coherent state with an average photon number @xmath103 , and the interplay between it and the atoms takes place in the buck - sukumar regime ( @xmath102 ) .
, title="fig:",width=287,height=188 ] ) plotted , separately , as a function of @xmath147 ( frame ( a ) ) and @xmath197 ( frame ( b ) ) for the time instants @xmath193 ( green ) , @xmath194 ( red ) , and @xmath195 ( blue ) .
the field is initially in a coherent state with an average photon number @xmath103 , and the interplay between it and the atoms takes place in the buck - sukumar regime ( @xmath102 ) .
, title="fig:",width=287,height=188 ] .
the parameters are @xmath190 ( blue line ) and @xmath199 ( red line ) , with @xmath102 . ,
width=340,height=207 ] in order to close and complement the present subsection , let us report briefly the influence of the nonlinearities of the model on the degree of entanglement between the two atoms .
this quantity can be suitably measured by dint of concurrence @xcite , which , for an arbitrary two - qubit system , is given by the expression @xmath200 , where the @xmath201 s are the square roots of the eigenvalues of the matrix @xmath202 in a non - increasing order , with @xmath179 being the reduced density matrix of the atomic system .
this calculation is entirely carried out numerically by solving the problem in the computational basis @xmath203 .
figure [ fig : concurrence ] displays the concurrence dynamics of the atoms for the same set of parameters as in fig . [
fig : purity - cs ] ( b ) for purity . from the figure
, one can see that the buck - sukumar coupling plays a preponderant role in bolstering the entanglement at certain times ( see blue curve ) that seem to be held approximately at integer multiples of @xmath204 within the shown interval .
this is in agreement with some results encountered for the evolution of concurrence using a non - interacting qubit as a _ probe _
@xcite and it is a consequence of the periodicity of the dynamics generated by the buck - sukumar interaction . a peculiar aspect that deserves to be noted is the fact that , on time domain , the appearance of the maxima of entanglement partially coincides with the one observed in purity dynamics just at such time instants ; this is so inasmuch as the latter was found to reach its maximum values at integer multiples of @xmath193 , approximately , at least within the time interval pointed out in fig .
[ fig : purity - cs ] ( b ) .
the presence of a kerr - like medium , on the other hand , tends to undermine the entanglement between the atoms ( red curve ) .
this conduct is also somewhat similar to the one observed for purity and becomes more and more unpredictable as time elapses and with the increase of the ratio of the anharmonicity parameter to the coupling constant @xmath197 .
once again , the dipole - dipole and ising atomic interactions did not give rise to significant changes in the foregoing evolution in the range of values @xmath205 and under the present nonlinear atom - field coupling scheme .
in this last section we will try to describe the influence of the nonlinearities of the model upon the evolution of the cavity field by analyzing it in terms of its entropy and representation on phase space .
let us first examine the degree of mixture of the state of the field in terms of the von neumann entropy @xcite @xmath206 where @xmath207 is the density operator of the system it seeks to describe , in our case the cavity field .
one way of facilitating the evaluation of this quantity is to invoke the theorem of araki and lieb @xcite from which it is stated that if the composite atom - field system is initially in a pure state , then the entropies of the field and atomic subsystems would have to have equal values at @xmath208 .
so , instead of using the reduced density matrix of the field to evaluate its entropy , one can equivalently make use of the reduced operator associated with the atomic system alone ( see eq .
( [ eq : rhoatoms ] ) ) , for which , by virtue of its dimension , we just have a set of three eigenvalues @xmath209 , and then the field entropy acquires the simpler form @xmath210 which ranges from 0 , for a pure state , to @xmath211 , for a maximally mixed state .
[ fig : entropy1 ] we show the evolution of the field entropy within the interval @xmath212 , where the field is initially in a coherent state with @xmath213 photons , and the initial state of atomic system is such that the two atoms are considered to be in the excited state @xmath214 ( fig .
[ fig : entropy1 ] ( a ) ) or the symmetric state @xmath215 ( fig .
[ fig : entropy1 ] ( b ) ) . in the figures ,
the black curves correspond to the case in which we have a linear coupling between the atoms and the field ( @xmath99 ) , and the cavity is supposed to be filled with a kerr - like medium such that @xmath216 .
the gray curves , on the other hand , describe the evolution of entropy when considering a buck - sukumar coupling , @xmath102 , in the case of a standard cavity ( @xmath89 ) .
besides begin highly dependent upon the initial state of the atomic system , the general behavior of the field entropy seems unpredictable when both the linear atom - field coupling and the nonlinear medium are included , save for the appearance of slightly pronounced and barely regular spikes at certain time instants ( see fig .
[ fig : entropy1 ] ( b ) , black curve ) indicating a tendency for the state of the field to recover , not completely , its purity when the atomic state is the maximally entangled one at the outset of its evolution .
this last behavior is bolstered by the nonlinear buck - sukumar atom - field coupling within a standard cavity during the first stages of the time evolution ( gray curves ) . during the shown time interval ,
there are specific time instants , integer multiples of @xmath217 in the dimensionless time variable @xmath218 , when the minima of the field entropy take place . at such time instants ,
the state of the field is almost pure when considering the state of the atoms to be initially in the maximally entangled state , a behavior that seems to be also in agreement with the atomic purity in fig .
[ fig : purity - cs ] ( b ) as expected . as a function of the whole atomic and anharmonicity parameters , @xmath147 and @xmath197 , respectively , and in a wider range of their values than the one considered so far , it is depicted in fig .
[ fig : entropy2 ] the field entropy at the time instants @xmath193 , @xmath194 and @xmath195 for the nonlinear coupling case described in fig .
[ fig : entropy1 ] ( b ) , gray curve , when the entropy seems to reach the first three minimum values . in similarity to the atomic purity ,
neither the dipole - dipole nor the ising interaction has significant bearing on the entropy of the field , which means that such interactions preserve to some extent the purity of the subsystems . on the other hand , at the same time instants
, the field evolves into a maximally mixed state quite rapidly with the increase of the anharmonicity parameter of the nonlinear medium , as seen in fig .
[ fig : entropy2 ] ( b ) , within the interval @xmath219 , approximately . for @xmath220 , we see once again a tendency to evolve into a pure state such as that of the atomic purity displayed in fig . [
fig : purity - cs2 ] ( b ) in this strong nonlinear regime .
+ ) when considering the atomic system being initially in its excited ( frame ( a ) ) and maximally entangled state ( frame ( b ) ) for a standard ( gray curve ) and a kerr - like cavity ( black curve ) .
the field starts being a coherent state with @xmath103 photons and its coupling with the atoms is nonlinear in buck - sukumar s sense ( @xmath102 ) and in the regime when @xmath221.,title="fig:",width=287,height=188 ] ) when considering the atomic system being initially in its excited ( frame ( a ) ) and maximally entangled state ( frame ( b ) ) for a standard ( gray curve ) and a kerr - like cavity ( black curve ) .
the field starts being a coherent state with @xmath103 photons and its coupling with the atoms is nonlinear in buck - sukumar s sense ( @xmath102 ) and in the regime when @xmath221.,title="fig:",width=287,height=188 ] ) and plotted , separately , as a function of @xmath147 ( frame ( a ) ) and @xmath197 ( frame ( b ) ) for the time instants @xmath193 ( green ) , @xmath194 ( red ) , and @xmath195 ( blue ) .
the field is initially in a coherent state with an average photon number @xmath103 , and the interplay between it and the atoms takes place in the buck - sukumar regime ( @xmath102 ) .
, title="fig:",width=287,height=188 ] ) and plotted , separately , as a function of @xmath147 ( frame ( a ) ) and @xmath197 ( frame ( b ) ) for the time instants @xmath193 ( green ) , @xmath194 ( red ) , and @xmath195 ( blue ) .
the field is initially in a coherent state with an average photon number @xmath103 , and the interplay between it and the atoms takes place in the buck - sukumar regime ( @xmath102 ) .
, title="fig:",width=287,height=188 ] let us now make a qualitative connection of the above results on entropy with a pictorial image provided by the so - called q function . before doing so ,
let us first establish the reduced density operator of the field computed via the following operation @xmath222 where @xmath223 now indicates the trace over the atomic variables , and @xmath224 , as stated earlier , denotes the density operator of the whole system composed of the field and the two atoms .
again , we shall regard the two atoms as being initially in their corresponding excited states , @xmath214 , or in the maximally entangled state , @xmath215 .
so , starting from the density operator of the composite system given by eq .
( [ eq : densityaf ] ) , together with the use of ( [ eq : tracerhoa ] ) , the reduced density operator of the field looks like @xmath225 and the time - dependent coefficient @xmath226 is chosen to be either the one given by eq .
( [ eq : coefficient ] ) or eq .
( [ eq : coefficient2 ] ) , depending upon the initial state of the two atoms . by definition ,
the husimi - q distribution is given by @xmath227 where , on substituting ( [ eq : rhof ] ) into the last expression , we thus arrive at the desired result @xmath228 so , with the help of the above expression for the q function , some interesting features of the radiation field in phase space will be described by focusing our attention primarily upon the role played by the nonlinear character of the model . in what follows all the results concerning the evolution on phase space
are presented by taking the cavity field to be initially in a coherent state with average photon number @xmath213 and the atoms to be in the non - interacting regime , i.e. , @xmath105 ; this is so because the contribution of the atomic interplay to the field s evolution has proven so far to be subtle at least within the restricted range of values of the effective atomic parameter @xmath147 and over the time scales considered here .
+ the corresponding results concerning the buck - sukumar regime , @xmath102 , and the cases in which the field is allowed to evolve within a standard or a nonlinear kerr - like cavity are shown in figs [ fig : qfunction1 ] and [ fig : qfunction2 ] , respectively , at times @xmath229 and @xmath195 , just when the entropy of the field described earlier reaches approximately some of its first minimum values , which was particularly noticeable when a standard cavity is taken into consideration . in this regard , we see in fig .
[ fig : qfunction1 ] that the development of the field turns out to be highly dependent upon the initial sate of the atomic system .
for instance , if the two atoms are initially placed in their excited state ( upper row ) , this circumstance can produce the formation of what seems to be well localized phase components of the field , from two to three of them at times @xmath193 , @xmath194 and @xmath230 , that are reminiscent of the so - called schrdinger - cat states , whereas only up to two phase components can be generated when the atoms are initially in the maximally entangled state ( lower row , at times @xmath193 and @xmath231 ) . however , based on the results of the field s entropy ( see fig . [
fig : entropy1 ] ( a ) , gray curve ) at such time instants , neither the triple nor the double phase components are altogether pure states of the field , given that the minimum of entropy is not small enough to guarantee the purity of such states , thus having certain properties of a mixed state to some degree .
this is not so when the field is almost re - focused into a single component , at @xmath232 , as it started off at the outset of its evolution .
on the other hand , if we now see fig .
[ fig : entropy1 ] ( b ) , gray curve , as a reference , it is found that the state of the field characterized by the double phase components at @xmath193 and @xmath230 , as well as the single ones at @xmath204 and @xmath195 , are almost pure states as far as the entropy s outcome is concerned .
this conduct repeats itself periodically during the very first stages of field s evolution and its representation is constrained to develop around the phase space origin on a circle of radius @xmath233 . + as for the kerr - like cavity , it is found that the nonlinear medium is able to create a richer cat - like splitting than the one we observe in the standard cavity case , as seen in some of the frames of fig .
[ fig : qfunction2 ] , say , at times @xmath234 and/or @xmath195 , depending on the initial state of the atomic system .
but , according to entropy s outcome at such time instants ( see , as a reference , figs .
[ fig : entropy1 ] ( a ) and ( b ) , black curve ) , the degree of purity of the corresponding field states is not - so - significant as is the previous case ; instead , most of the time they exhibit the proper character of a mixed state regardless of the initial state of the atoms , and their frequency of appearance is no longer periodic as a function of time . at the beginning of the evolution ,
the kerr effect manifests itself in the distortion of the field , namely , in some cases , the phase of it is warped and curved in a single or double deformed elongated rim at first ( see , for instance , frame ( a ) , upper row ; frames ( e ) and ( f ) , lower row ) and then is spread over a kind of phase space ring composed of a number of cat - like states as time elapses ( frame ( g ) , lower row ) .
this behavior was proven to be more unpredictable throughout an extended time interval far beyond the one pointed out in the present results .
finally , it is worth mentioning that the kerr effect on field s phase space has thoroughly been investigated by several authors in several circumstances involving atom - field interactions , among them we can cite the work of werner and risken @xcite on the dynamics of the q - function for a two - level atom ; the generation of superpositions of distinguishable states by means of yurke and stoler s model @xcite ; the treatment of one- and two - photon jaynes - cummings models with one atom via a pegg - barnett hermitian phase operator formalism undertaken by gantsog @xcite ; and , more recently , a nonlinear version of the jaynes - cummings model and its nonlinear coherent states based upon the f - deformed oscillator formalism has also been introduced @xcite .
photons and the atomic system in its excited ( upper row ) or maximally entangled state ( lowger row ) for the time instants @xmath235 , and @xmath195 . within the buck - sukumar regime , @xmath102 ,
the plots are computed by considering the following parameters : @xmath96 , @xmath221 ( non - interacting atoms ) , and @xmath236 ( standard cavity).,title="fig:",width=642,height=151 ] photons and the atomic system in its excited ( upper row ) or maximally entangled state ( lowger row ) for the time instants @xmath235 , and @xmath195 . within the buck - sukumar regime , @xmath102 , the plots are computed by considering the following parameters : @xmath96 , @xmath221 ( non - interacting atoms ) , and @xmath236 ( standard cavity).,title="fig:",width=642,height=151 ] photons and the atomic system in its excited ( upper row ) or maximally entangled state ( lowger row ) for the time instants @xmath235 , and @xmath195 . within the buck - sukumar regime , @xmath102 ,
the plots are computed by considering the following parameters : @xmath96 , @xmath221 ( non - interacting atoms ) , and @xmath216 ( kerr - like cavity).,title="fig:",width=642,height=151 ] photons and the atomic system in its excited ( upper row ) or maximally entangled state ( lowger row ) for the time instants @xmath235 , and @xmath195 . within the buck - sukumar regime , @xmath102 ,
the plots are computed by considering the following parameters : @xmath96 , @xmath221 ( non - interacting atoms ) , and @xmath216 ( kerr - like cavity).,title="fig:",width=642,height=151 ]
in this work we have put forward and provided the solution of a nonlinear version of the jaynes - cummings model for two identical two - level atoms allowing for ising - like and dipole - dipole interactions between them . the model is nonlinear in a twofold sense , that is , it incorporates the possibility of describing the evolution of the radiation field in a cavity that can be supposed to be filled with an nonlinear medium and/or a general intensity - dependent coupling between the field and the two atoms , each of these characteristics being specified , separately , by the proper photon - number - dependent function .
even though we have considered two two - level atoms interacting ( nonlinearly ) with a quantized field , we emphasize the fact that such nonlinearities may appear in analogous systems ( in the sense that hamiltonians like the one we consider here may be engineered ) such as trapped ions interacting with classical laser fields @xcite .
if , for instance , we consider two trapped ions , close enough together to interact between them , as represented in ( [ eq : haa ] ) , and a set of lasers detuned correctly , one can generate nonlinearities that behave as associated laguerre polynomials , and by adding properly , one may easily reproduce a number of possible nonlinear models we are familiar with from nonlinear optics , such as the usual kerr - type and buck - sukumar nonlinearities or even more complicated coupling schemes . therefore , our treatment is not only valid for the interaction between a quantized field and two two - level atoms , but also for other ( algebraically ) similar systems .
+ some dynamical properties of the atomic and field subsystems , such as the atomic excitation , purity and concurrence , as well as the entropy and phase space dynamics of the field , have been described by considering a kerr - like cavity and/or a buck - sukumar atom - field coupling , as particular examples , in the resonant quantum dynamics ( @xmath96 ) .
the most interesting outcomes of the model come from featuring such nonlinearities or merging any of them with the atomic interplay under certain circumstances . in this regard , it was found that the combination of the nonlinear atom - field coupling and the dipole - dipole - and - ising - like atomic interaction , in the range of values @xmath237 , or merging the former with the anharmonic effect of a kerr - type cavity , within the regime @xmath238 , gives rise to the appearance of a noticeable beating pattern and the inversion of successive revivals in the evolution of atomic excitation , and from the algebraic results we have been able to obtain an approximate closed expression for the said property that has allowed us to estimate the origin and duration of such beats , as well as the collapse and revival times .
both the purity and concurrence atomic properties are bolstered by the buck - sukumar coupling itself at certain time instants , and by the kerr - like medium up to some degree , particularly when the atomic system is in its maximally entangled state . as to the atomic purity and field entropy , it was also found that the dipole - dipole and ising effects , though not - so - significant by themselves , become manifest in the regime when the ratio of the whole atomic parameter to the constant coupling is such that @xmath239 , but nevertheless we suggest that one should be careful of working on this regime given that , as far as we know , it has not been proven in the literature to be suitably justified at such high values ; this is why we have chosen here to be moderate in the use of small values of the aforesaid parameter .
finally , we have explored the dynamics of the field in terms of its entropy and evolution on phase space via the q - function , where the role played by buck - sukumar nonlinear coupling and the kerr medium were found to be preponderant in the splitting of phase properties of the field giving rise to schrdinger - cat - like states .
we aknowledge partial support from projects dgapa unam in113016 and conacyt 166961 .
one of the authors ( o de los santos - snchez ) wishes to thank the _ red temtica de tecnologas cunticas _ for financial support through project 250785 conacyt - cnpq mxico .
cgg is grateful to conacyt for financial support under doctoral fellowship no .
last but not least , we especially thank both professor h. m. moya - cessa and an anonymous referee for helpful suggestions that greatly improved the content of the paper .
99 jaynes e t and cummings f w 1963 _ proc . ieee _ * 51 * 89 shore b w and knight p l 1993 _ j. mod
_ * 40 * 1195 joshi a and xiao m 2004 _ j. opt .
soc . am . _
b * 21 * 1685 eberly j h , narozhny n b and snchez - mondragon j j 1980 _ phys .
lett . _ * 44 * 1323 parker j and stroud c r 1986 _ j. opt .
b * 3 * 96 parkins a s 1990 _ phys .
a , * 42 * 6873 joshi a and puri r r 1990 _ phys . rev . _ a * 42 * 4346 buck b and sukumar c v 1981 _ phys .
lett . _ * 81 * 132 kochetov e a 1987 _ j. phys .
gen . _ * 20 * 2433 cordeiro f , provid^ encia c , providncia j da and nishiyama s 2007 _ j. phys .
a : math . theor .
_ * 40 * 12153 penna v and raffa f a 2014 _ int .
j. quantum inform . _
* 12 * 1560010 napoli a and messina a 1996 _ j. mod . opt . _ * 43 * 649 napoli a and messina a 1996 _ nucl .
instr . and meth .
_ b * 116 * 465 buzek v and jex i 1990 _ opt .
commun . _ * 78 * 425 joshi a and puri r r 1992 _ phys .
rev . _ a * 45 * 5056 berlin g and aliaga j 2001 _ j. mod .
opt . _ * 48 * 1819 cordero s and rcamier j 2011 _ j. phys .
b : at . mol .
phys . _ * 44 * 135502 deng z 1985 _ opt .
commun . _ * 54 * 222 joshi a , puri r r and lawande s v 1991 _ phys .
rev . _ a * 44 * 2135 zhi - ming zhang , lei xu and zhen - fei luo 1993 _ phys .
rev . _ a * 47 * 1557 chilingaryan s a , rodrguez - lara b m 2013 _ j. phys .
a _ * 46 * 335301 tavis m and cummings f w 1968 _ phys .
rev . _ * 170 * 279 tavis m and cummings f w 1969 _ phys . rev . _
* 188 * 692 torres j m , sadurn e and seligman t h 2010 _ j. phys .
theor . _ * 43 * 192002 amaro j g and pineda c 2014 _ phys .
_ * 90 * 068019 sorensen a and molmer k 1999 _ phys .
* 83 * 2274 porras d and cirac j i 2004 _ phys .
_ * 92 * 207901 hartmann m j , brando g s l and plenio m b 2007 _ phys .
lett . _ * 99 * 160501 cusati t , napoli a and messina a 2006 _ journal of molecular structure : theochem _ * 769 * 3 napoli a , messina a , cusati t and draganescu g 2006 _ eur . phys . j. _
b * 50 * 419 torres j m , bernd j z and alber g 2016 _ appl . phys .
b _ * 122 * 1 crespi a , longhi s and osellame r 2012 _ phys .
lett . _ * 108 * 163601 rodrguez - lara b m , soto - eguibar f , zrate crdenas a and moya - cessa h m 2013 _ opt
. express _ * 21 * 12888 rodrguez - lara b m 2014 _ j. opt
b * 31 * 1719 blais a , huang ren - shou , wallraff a , girvin s m and schoelkopf 2004 _ phys
. rev . _ a * 69 * 062320 wallentowitz s and vogel w 1997 _ phys .
rev . _ a * 55 * 4438 de matos
filho r l and vogel w 1998 _ phys .
rev . _ a * 58 * r1661 manko v , marmo g , porzio a , solimeno s and zaccaria f 2000 _ phys .
rev . _ a * 62 * 053407 - 1 leibfried d , blatt r , monroe c and wineland d 2003 _ rev .
phys . _ * 75 * 281 moya - cessa h m , soto - eguibar f , vargas - martnez j m , jurez - amaro r and ziga - segundo a 2012 _ phys .
rep . _ * 513 * 229 mandel l and wolf e 1995 _ optical coherence and quantum optics _
( cambridge university press ) p 1102 agarwal g s and puri r r 1989 _ phys .
a * 39 * 2969 mancini s and tombesi p 1995 _ phys .
_ a * 52 * 2475 yoo h i and eberly j h 1985 _ phys .
_ * 118 * 239 averbuch i s and perelman n f 1989 _ phys .
_ a , * 139 * , 449 gra p and jedrzejek c 1992 _ phys . rev . _
a * 45 * 6816 wootters w k 1998 _ phys . rev . lett .
_ * 80 * 2245 gonzlez - gutirrez c , romn - ancheyta r , espitia d , and lo franco r 2016 _ arxiv preprint arxiv:1604.04671 _ nielsen m a and
chuang i l 2000 _ quantum computation and quantum information _ ( cambridge university press ) p 510 araki h and lieb e h 1970 _ commun . math .
phys . _ * 18 * 160 werner m j and risken h 1991 _ quantum opt . _
* 3 * 185 yurke b and stoler d 1986 _ phys .
lett . _ * 57 * 13 gantsog ts , joshi a and tanas r 1996 _ quantum semiclass
_ * 8 * 445 de los santos - snchez o and rcamier j 2012 _ j. phys .
b : at . mol .
* 45 * 015502 | in this work we examine a nonlinear version of the jaynes - cummings model for two identical two - level atoms allowing for ising - like and dipole - dipole interplays between them .
the model is said to be nonlinear in the sense that it can incorporate both a general intensity - dependent interaction between the atomic system and the cavity field and/or the presence of a nonlinear medium inside the cavity . as an example
, we consider a particular type of atom - field coupling based upon the so - called buck - sukumar model and a lossless kerr - like cavity .
we describe the possible effects of such features on the evolution of some quantities of current interest , such as atomic excitation , purity , concurrence , the entropy of the field and the evolution of the latter in phase space . |
while abstract polytopes are a combinatorial generalisation of classical polyhedra and polytopes , maniplexes generalise maps on surfaces and ( the flag graph of ) abstract polytopes .
the combinatorial structure of maniplexes , maps and polytopes is completely determined by a edge - coloured @xmath2-valent graph with chromatic index @xmath2 , often called the flag graph .
the symmetry type graph of a map is the quotient of its flag graph under the action of the automorphism group . in this paper
we extend the notion of symmetry type graphs of maps to that of maniplexes ( and polytopes ) .
given a maniplex , its symmetry type graph encapsulates all the information of the local configuration of the flag orbits under the action of the automorphism group of the maniplex .
traditionally , the main focus of the study of maps and polytopes has been that of their symmetries .
regular and chiral ones have been extensively studied .
these are maps and polytopes with either maximum degree of symmetry or maximum degree of symmetry by rotation .
edge - transitive maps were studied in @xcite by siran , tucker and watkins .
such maps have either 1 , 2 or 4 orbits of flags under the action of the automorphism group .
more recently orbani , pellicer and weiss extend this study and classify @xmath0-orbit maps ( maps with @xmath0 orbits of flags under the automorphism group ) up to @xmath3 in @xcite .
little is known about polytopes that are neither regular nor chiral . in @xcite hubard
gives a complete characterisation of the automorphism group of 2-orbit and fully - transitive polyhedra ( i.e. polyhedra transitive on vertices , edges and faces ) in terms of distinguished generators of them .
moreover , she finds generators of the automorphism group of a 2-orbit polytope of any given rank .
symmetry type graphs of the platonic and archimedean solids were determined in @xcite . in @xcite
del ro - francos , hubard , orbani and pisanski determine symmetry type graphs of up to 5 vertices and give , for up to 7 vertices , the possible symmetry type graphs that a properly self - dual , an improperly self - dual and a medial map might have .
the possible symmetry type graphs that a truncation of a map can have is determined in @xcite .
one can find in @xcite a strategy to generate symmetry type graphs . by making use of symmetry type graphs , in this paper
we classify 3-orbit polytopes and give generators of their automorphism groups .
in particular , we show that 3-orbit polytopes are never fully - transitive , but they are @xmath4-face - transitive for all @xmath4 but one or two , depending on the class . we extend further the study of symmetry type graphs to show that if a 4-orbit polytope is not fully - transitive , then it is @xmath4-face - transitive for all @xmath4 but at most three ranks .
moreover , we show that a fully - transitive 3-maniplex ( or 4-polytope ) that is not regular can not have an odd number of orbits of flags , under the action of the automorphism group .
the main result of the paper is stated in theorem [ auto ] . given a maniplex @xmath5 in theorem [ auto ] we give generators for the automorphism group of @xmath5 with respect to some base flag .
the paper is divided into six sections , organised in the following way . in section [ sec : polymani ] , we review some basic theory of polytopes and maniplexes , and describe their respective flag graphs . in section [ sec : stg ] , we define and give some properties of the symmetry type graphs of polytopes and maniplexes , extending the concept of symmetry type graphs of maps . in section [ sec : stg - highly ] , we study symmetry type graphs of highly symmetric maniplexes . in particular , we classify symmetry type graphs with 3 vertices , determine the possible transitivities that a 4-orbit mainplex can have and study some properties of fully - transitive maniplexes of rank 3 . in section [ gen - autg ]
we give generators of the automorphism group of a polytope or a maniplex . in the last section of the paper we define oriented and orientable maniplexes .
further on , we define the oriented flag di - graph which emerge from a flag graph if this is bipartite . the oriented symmetry type di - graph of an oriented maniplex is then a quotient of the oriented flag di - graph , just as the symmetry type graph was a quotient of the flag graph . using these graphs we classify oriented 2-orbit maniplexes and give generators for their orientation preserving automorphism group .
in this section we briefly review the basic theory of abstract polytopes and their monodromy groups ( for details we refer the reader to @xcite and @xcite ) . an ( _ abstract _ ) _ polytope of rank _
@xmath2 , or simply an _ @xmath2-polytope _ , is a partially ordered set @xmath6 with a strictly monotone rank function with range @xmath7 .
an element of rank @xmath8 is called a _
@xmath8-face _ of @xmath6 , and a face of rank @xmath9 , @xmath10 or @xmath11 is called a _ vertex _ , _ edge _ or _ facet _ , respectively .
a _ chain of @xmath12 _ is a totally ordered subset of @xmath12 .
the maximal chains , or _ flags _ , all contain exactly @xmath13 faces , including a unique least face @xmath14 ( of rank @xmath15 ) and a unique greatest face @xmath16 ( of rank @xmath2 ) .
a polytope @xmath6 has the following homogeneity property ( diamond condition ) : whenever @xmath17 , with @xmath18 a @xmath19-face and @xmath20 a @xmath21-face for some @xmath8 , then there are exactly two @xmath8-faces @xmath22 with @xmath23 .
two flags are said to be _ adjacent _ ( @xmath4-_adjacent _ ) if they differ in a single face ( just their @xmath4-face , respectively ) .
the diamond condition can be rephrased by saying that every flag @xmath24 of @xmath12 has a unique @xmath4-adjacent flag , denoted @xmath25 , for each @xmath26 .
finally , @xmath6 is _ strongly flag - connected _ , in the sense that , if @xmath24 and @xmath27 are two flags , then they can be joined by a sequence of successively adjacent flags , each containing @xmath28 .
let @xmath6 be an abstract @xmath2-polytope .
universal _ string coxeter group @xmath29 $ ] of rank @xmath2 , with distinguished involutory generators @xmath30 @xmath31 , acts transitively on the set of flags @xmath32 of @xmath12 in such a way that @xmath33 , the @xmath4-adjacent flag of @xmath27 , for each @xmath34 and each @xmath27 in @xmath32 . in particular ,
if @xmath35 then @xmath36 the _ monodromy or connection group _ of @xmath12 ( see for example @xcite ) , denoted @xmath37 , is the quotient of @xmath38 by the normal subgroup @xmath39 of @xmath38 consisting of those elements of @xmath38 that act trivially on @xmath32 ( that is , fix every flag of @xmath40 .
let @xmath41 denote the canonical epimorphism . clearly , @xmath37 acts on @xmath32 in such a way that @xmath42 for each @xmath43 in @xmath38 and each @xmath27 in @xmath32 , so in particular @xmath44 for each @xmath4 .
we slightly abuse notation and also let @xmath45 denote the @xmath4-th generator @xmath46 of @xmath37 .
we shall refer to these @xmath45 as the _ distinguished _ generators of @xmath37 .
since the action of @xmath38 is transitive on the flags , the action of @xmath37 on the flags of @xmath12 is also transitive ; moreover , this action is faithful , since only the trivial element of @xmath37 fixes every flag .
thus @xmath37 can be viewed as a subgroup of the symmetric group on @xmath32 .
note that for every flag @xmath24 of @xmath12 and @xmath47 such that @xmath48 , we have that @xmath49 .
since the action of @xmath37 is faithful in @xmath32 , this implies that @xmath50 , whenever @xmath48 .
an _ automorphism _ of a polytope @xmath12 is a bijection of @xmath12 that preserves the order .
we shall denote by @xmath51 the group of automorphisms of @xmath12 .
note that any automorphism of @xmath12 induces a bijection of its flags that preserves the @xmath4-adjacencies , for every @xmath52 .
a polytope @xmath12 is said to be _ regular _ if the action of @xmath51 is regular on @xmath32 .
if @xmath51 has exactly 2 orbits on @xmath32 in such a way that adjacent flags belong to different orbits , @xmath12 is called a _
chiral polytope_. we say that a polytope is a _
@xmath0-orbit polytope _ if the action of @xmath51 has exactly @xmath0 orbits on @xmath32 .
hence , regular polytopes are 1-orbit polytopes and chiral polytopes are 2-orbit polytopes . given an @xmath2-polytope @xmath12 , we define the _ graph of flags _ @xmath53 of @xmath12 as follows .
the vertices of @xmath53 are the flags of @xmath12 , and we put an edge between two of them whenever the corresponding flags are adjacent . hence @xmath53 is @xmath2-valent ( i.e. every vertex of @xmath53 has exactly @xmath2 incident edges ; to reduce confusion we avoid the alternative terminology ` @xmath2-regular ' ) .
furthermore , we can colour the edges of @xmath53 with @xmath2 different colours as determined by the adjacencies of the flags of @xmath12 . that is , an edge of @xmath53 has colour @xmath4 , if the corresponding flags of @xmath12 are @xmath4-adjacent . in this way
every vertex of @xmath53 has exactly one edge of each colour ( see figure [ fig : baricentic ] ) .
it is straightforward to see that each automorphism of @xmath12 induces an automorphism of the flag graph @xmath53 that preserves the colours .
conversely , every automorphism of @xmath53 that preserves the colours is a bijection of the flags that preserves all the adjacencies , inducing an automorphism of @xmath12 .
that is , the automorphism group @xmath51 of @xmath12 is the colour preserving automorphism group @xmath54 of @xmath53 .
note that the connectivity of @xmath12 implies that the action of @xmath51 on @xmath32 is free ( or semiregular ) .
hence , the action of @xmath54 is free on the vertices of the graph @xmath53 .
one can re - label the edges of @xmath53 and assign to them the generators of @xmath37 .
in fact , since for each flag @xmath24 , the action of @xmath45 takes @xmath24 to @xmath55 , by thinking of the edge of colour @xmath4 of @xmath53 as the generator @xmath45 , one can regard a walk along the edges of @xmath53 as an element of @xmath37 .
that is , if @xmath43 is a walk along the edges of @xmath37 that starts at @xmath24 and finishes at @xmath27 , then we have that @xmath56 .
hence , the connectivity of @xmath12 also implies that the action of @xmath37 is transitive on the vertices of @xmath53 .
furthermore , since the @xmath4-faces of @xmath12 can be regarded as the orbits of flags under the action of the subgroup @xmath57 , the @xmath4-faces of @xmath12 can be also regarded as the connected components of the subgraph of @xmath53 obtained by deleting all the edges of colour @xmath4 .
maniplexes were first introduced by steve wilson in @xcite , aiming to unify the notion of maps and polytopes . in this section
we review the basic theory of them .
an _ @xmath2-complex _
@xmath58 is defined by a set of _ flags _ @xmath59 and a sequence @xmath60 @xmath61 , such that each @xmath45 partitions the set @xmath59 into sets of size 2 and the partitions defined by @xmath45 and @xmath62 are disjoint when @xmath63 .
furthermore , we ask for @xmath5 to be _ connected _ in the following way .
thinking of the @xmath2-complex @xmath58 as the graph @xmath64 with vertex set @xmath59 , and with edges of colour @xmath4 corresponding to the matching @xmath45 , we ask for the graph @xmath64 indexed by @xmath5 to be connected .
an _ @xmath2-maniplex _ is an @xmath2-complex such that the elements in the sequence @xmath65 correspond to the distinguished involutory generators of a coxeter string group . in terms of the graph @xmath64
, this means that the connected components of the induced subgraph with edges of colours @xmath4 and @xmath8 , with @xmath66 are 4-cyles
. we shall refer to the _ rank _ of an @xmath2-maniplex , precisely to @xmath2 .
a 0-maniplex must be a graph with two vertices joined by an edge of colour 0 .
a 1-maniplex is associated to a 2-polytope or @xmath67-gon , which graph contains @xmath68 vertices joined by a perfect matching of colours 0 and 1 , and each of size @xmath67 .
a 2-maniplex can be considered as a map and vice versa , so that maniplexes generalise the notion of maps to higher rank . regarding polytopes
, the flag graph of any @xmath69-polytope can be associated to an @xmath2-maniplex , generalising in such way the notion of polytopes .
one can think of the sequence @xmath65 of a maniplex @xmath5 as permutations of the flags .
in fact , if @xmath70 are flags of @xmath5 belonging to the same part of the partition induced by @xmath45 , for some @xmath4 , we say that @xmath71 and @xmath72 . in this way
each @xmath45 acts as a involutory permutation of @xmath59 . in analogy with polytopes , we let @xmath73 and define the _ connection group _ @xmath74 of @xmath5 as the quotient of @xmath75 over @xmath39 . as before , we abuse notation and say that @xmath74 is generated by @xmath76 and define the action of @xmath74 on the flags inductively , induced by the action of the sequence @xmath65 . in this way , the action of @xmath74 on @xmath59 is faithful and transitive . note
further that since the sequence @xmath65 induces a string coxeter group , then , as elements of @xmath74 , @xmath50 whenever @xmath48 .
this implies that given a flag @xmath24 of @xmath5 and @xmath77 such that @xmath48 , we have that @xmath78 .
an _ automorphism _ @xmath79 of an @xmath2-maniplex is a colour - preserving automorphism of the graph @xmath64 . in a similar way
as it happens for polytopes , the connectivity of the graph @xmath80 implies that the action of the automorphism group @xmath81 ) of @xmath5 is free on the vertices of @xmath80 .
hence , @xmath79 can be seen as a permutation of the flags in @xmath59 that commutes with each of the permutations in the connection group . to have consistent concepts and notation between polytopes and maniplexes , we shall say that an @xmath4-face ( or a face of rank @xmath4 ) of a maniplex is a connected component of the subgraph of @xmath80 obtained by removing the @xmath4-edges of @xmath80 .
furthermore , we say that two flags @xmath24 and @xmath27 are @xmath4-adjacent if @xmath82 ( note that since @xmath45 is an involution , @xmath82 implies that @xmath83 , so the concept is symmetric ) . to each @xmath4-face @xmath18 of @xmath5 , we can associate an @xmath84-maniplex @xmath85 by identifying two flags of @xmath18 whenever there is a @xmath8-edge between them , with @xmath86 .
equivalently , we can remove from @xmath18 all edges of colours @xmath87 , and then take one of the connected components .
in fact , since @xmath88 commutes with @xmath89 , the connected components of this subgraph of @xmath18 are all isomorphic , so it does not matter which one we pick .
if @xmath24 is a flag of @xmath5 that contains the @xmath4-face @xmath18 , then it naturally induces a flag @xmath90 in @xmath85 .
similarly , if @xmath91 fixes @xmath18 , then @xmath92 induces an automorphism @xmath93 , defined by @xmath94 . to check that this is well - defined , suppose that @xmath95 ; we want to show that @xmath96 . since @xmath95 , it follows that @xmath97 for some @xmath98 .
then @xmath99 , so that @xmath100 . by definition ,
the edges of @xmath80 of one given colour form a perfect matching .
the 2-factors of the graph @xmath80 are the subgraphs spanned by the edges of two different colours of edges . since the automorphisms of @xmath5 preserve the adjacencies between the flags , it is not difficult to see that the following lemma holds .
[ orbittoorbit ] let @xmath24 be a flag of @xmath5 and let @xmath101 .
if @xmath102 and @xmath103 denote the flag orbits of @xmath24 and @xmath104 ( under @xmath105 ) , respectively , then @xmath106
if and only if @xmath107 . we say that a maniplex @xmath5 is _ @xmath4-face - transitive _ if @xmath105 is transitive on the faces of rank @xmath4 .
we say that @xmath5 is _ fully - transitive _ if it is @xmath4-face - transitive for every @xmath108 .
if @xmath105 has @xmath0 orbits on the flags of @xmath5 , we say that @xmath5 is a @xmath0-orbit maniplex .
a 1-orbit maniplex is also called a _
reflexible _ maniplex .
a 2-orbit maniplex with adjacent flags belonging to different orbits is a _
chiral _ maniplex .
if a maniplex has at most 2 orbits of flags and @xmath109 is a bipartite graph , then the maniplex is said to be _
in this section we shall define the symmetry type graph of a polytope or a maniplex . to this end
, we shall make use of quotient of graphs .
therefore , we now consider pregraphs ; that is , graphs that allow multiple edges and semi - edges .
as it should be clear , it makes no difference whether we consider an abstract @xmath2-polytope or an @xmath110-maniplex . hence , though we will consider maniplexes throughout the paper , similar results will apply to polytopes .
given an edge - coloured graph @xmath80 , and a partition @xmath111 of its vertex set @xmath112 , the _ coloured quotient with respect to @xmath111 _
, @xmath113 , is defined as the pregraph with vertex set @xmath111 , such that for any two vertices @xmath114 , there is a dart of colour @xmath115 from @xmath116 to @xmath117 if and only if there exists @xmath118 and @xmath119 such that there is a dart of colour @xmath115 from @xmath120 to @xmath121 . edges between vertices in the same part of the partition @xmath111 quotient into semi - edges . throughout the remainder of this section ,
let @xmath5 be an @xmath110-maniplex and @xmath109 its coloured flag graph . as we discussed in the previous section , @xmath105 acts semiregularly on the vertices of @xmath109 .
we shall consider the orbits of the vertices of @xmath109 under the action of @xmath105 as our partition @xmath122 , and denote @xmath123 .
note that since the action is semiregular , every two orbits @xmath124 have the same number of elements .
the _ symmetry type graph _
@xmath125 of @xmath5 is the coloured quotient graph of @xmath109 with respect to @xmath126 .
since the flag graph @xmath109 is an undirected graph , then @xmath125 is a pre - graph without loops or directed edges .
furthermore , as we are taking the coloured quotient , and @xmath109 is edge - coloured with @xmath2 colours , then @xmath125 is an @xmath2-valent pre - graph , with one edge or semi - edge of each colour at each vertex .
it is hence not difficult to see that if @xmath5 is a reflexible maniplex , then @xmath125 is a graph consisting of only one vertex and @xmath2 semi - edges , all of them of different colours .
in fact , the symmetry type graph of a @xmath0-orbit maniplex has precisely @xmath0 vertices .
figure [ stgrank3a ] shows the symmetry type graph of a reflexible 2-maniplex ( on the left ) , and the symmetry type graph of the cuboctahedron : the quotient graph of the flag graph in figure [ fig : baricentic ] with respect to the automorphism group of the cubocahedron .
note that by the definition of @xmath125 , there exists a surjective function @xmath127 that assigns , to each vertex of @xmath128 its corresponding orbit in @xmath125 .
hence , given @xmath129 , we have that @xmath130 if and only if @xmath24 and @xmath27 are in the same orbit under @xmath105 . given vertices @xmath131 of @xmath125 ,
if there is an @xmath4-edge joining them , we shall denote such edge as @xmath132 .
similarly , @xmath133 shall denote the semi - edge of colour @xmath4 incident to the vertex @xmath121 .
because of lemma [ orbittoorbit ] , we can define the action of @xmath74 on the vertices of @xmath125 .
in fact , given @xmath134 and @xmath101 , then @xmath135 , where @xmath136 . note that the definition of the action does not depend on the choice of @xmath136 ; in fact , we have that @xmath137 if and only if @xmath130 and this in turn is true if and only if @xmath24 and @xmath27 are in the same orbit under @xmath105 . by lemma [ orbittoorbit ] ,
the fact that @xmath24 and @xmath27 are in the same orbit under @xmath105 implies that , for any @xmath101 , the flags @xmath104 and @xmath138 are also in the same orbit under @xmath105 .
hence @xmath139 and therefore the definition of @xmath140 does not depend on the choice of the element @xmath136 . since @xmath74 is transitive on the vertices of @xmath109 , then it is also transitive on the vertices of @xmath125 , implying that @xmath125 is a connected graph .
furthermore , the action of each generator @xmath45 of @xmath74 on a vertex @xmath121 of @xmath125 corresponds precisely to the ( semi-)edge of colour @xmath4 incident to @xmath121 .
hence , the orbit @xmath141 corresponds to the orbit under @xmath105 of an @xmath4-face @xmath18 of @xmath5 such that @xmath142 , for some @xmath143 ( as before , different choices of flag @xmath144 induce the same orbit of @xmath4-faces ) .
therefore , the connected components of the subgraph @xmath145 of @xmath125 with edges of colours @xmath146 correspond to the orbits of the @xmath4-faces under @xmath105 . in particular
this implies the following proposition .
let @xmath5 be a maniplex , @xmath125 its symmetry type graph and let @xmath145 be the subgraph resulting by erasing the @xmath4-edges of @xmath125 .
then @xmath5 is @xmath4-face - transitive if and only if @xmath145 is connected .
we shall say that a symmetry type graph @xmath147 is @xmath4-face - transitive if @xmath148 is connected , and that @xmath147 is a fully - transitive symmetry type graph if it is @xmath4-face - transitive for all @xmath4 .
recall that to each @xmath4-face @xmath18 of @xmath5 , there is an associated @xmath84-maniplex @xmath85 .
the symmetry type graph @xmath149 is related in a natural way to the connected component of @xmath145 that corresponds to @xmath18 : [ stgoffaces ] let @xmath18 be an @xmath4-face of the maniplex @xmath5 , and let @xmath85 be the corresponding @xmath84-maniplex .
let @xmath150 be the connected component of @xmath145 corresponding to @xmath18 .
then there is a surjective function @xmath151 .
furthermore , if @xmath152 then each @xmath8-edge @xmath153 of @xmath150 yields a @xmath8-edge @xmath154 in @xmath149 , and if @xmath86 , then @xmath155 .
first , let @xmath24 and @xmath27 be flags of @xmath5 that are both in the connected component @xmath18 , and suppose that they lie in the same flag orbit , so that @xmath156 for some @xmath91 .
then the induced automorphism @xmath157 of @xmath85 sends @xmath90 to @xmath158 , and therefore @xmath90 and @xmath158 lie in the same orbit .
furthermore , every flag of @xmath85 is of the form @xmath90 for some @xmath24 in @xmath18 .
thus , each orbit of @xmath5 that intersects @xmath18 induces an orbit of @xmath85 , and it follows that there is a surjective function @xmath151 . consider an edge @xmath153 in @xmath150 .
then @xmath159 for some flag @xmath24 in @xmath18 , and we can take @xmath160 . both @xmath24 and @xmath161 induce flags in @xmath85 . if @xmath152 , then @xmath162 .
therefore , there must be a @xmath8-edge from the orbit of @xmath90 to the orbit of @xmath163 ; in other words , a @xmath8-edge from @xmath164 to @xmath165 . on the other hand ,
if @xmath86 , then @xmath166 , and so @xmath90 and @xmath163 lie in the same orbit and thus @xmath167 .
note that the edges of a given colour @xmath4 of @xmath125 form a perfect matching ( where , of course , we are allowing to match a vertex with itself by a semi - edge ) . given two colours @xmath4 and @xmath8 , the subgraph of @xmath125 consisting of all the vertices of @xmath125 and only the @xmath4- and @xmath8-edges
shall be called a @xmath168 2-factor of @xmath125 . because @xmath50 whenever @xmath169 , in @xmath109 , the alternating cycles of colours @xmath4 and @xmath8 have length 4 . by lemma [ orbittoorbit ]
each of these 4-cycles should then factor , in @xmath125 , into one of the five graphs in figure [ 4cyclequotient ] .
hence , if @xmath169 , then the connected components of the @xmath168 2-factors of @xmath125 are precisely one of these graphs .
coloured 4-cycles.,width=302 ] in light of the above observations we state the following lemma .
[ 2factors4vertices ] let @xmath125 be the symmetry type graph of a maniplex . if there are vertices @xmath170 such that @xmath171 with @xmath66 , then the connected component of the @xmath168 2-factor that contains @xmath121 has four vertices
one can classify maniplexes with small number of flag orbits ( under the action of the automorphism group of the maniplex ) in terms of their symmetry type graphs .
the number of distinct possible symmetry types of a @xmath0-orbit @xmath110-maniplex is the number of connected pre - graphs on @xmath0 vertices that are @xmath2-valent and that can be edge - coloured with exactly @xmath2 colours .
furthermore , given a symmetry type graph , one can read from the appropriate coloured subgraphs the different types of face transitivities that the maniplex has . as pointed out before , the symmetry type graph of a reflexible @xmath110-maniplex consists of one vertex and @xmath2 semi - edges .
the classification of two - orbit maniplexes ( see @xcite ) in terms of the local configuration of their flags follows immediately from considering symmetry type graphs .
in fact , for each @xmath2 , there are @xmath172 symmetry type graphs with 2 vertices and @xmath2 ( semi)-edges , since given any proper subset @xmath173 of the colours @xmath174 , there is a symmetry type graph with two vertices , @xmath175 semi - edges corresponding to the colours of @xmath173 , and where all the edges between the two vertices use the colours not in @xmath173 ( see figure [ 2orbitstg ] ) .
this symmetry type graph corresponds precisely to polytopes in class @xmath176 , see @xcite .
.,width=302 ] highly symmetric maniplexes can be regarded as those with few flag orbits or those with many ( or all ) face transitivities . in @xcite
one can find the complete list of symmetry type graphs of @xmath177-maniplexes with at most 5 vertices . in this section
we classify symmetry type graphs with 3 vertices and study some properties of symmetry type graphs of 4-orbit maniplexes and fully - transitive 3-maniplexes .
[ stg_3-orbit ] there are exactly @xmath178 different possible symmetry type graphs of 3-orbit maniplexes of rank @xmath11 .
let @xmath5 be a 3-orbit @xmath110-maniplex and @xmath125 its symmetry type graph .
then , @xmath125 is an @xmath2-valent well edge - coloured graph with vertices @xmath179 and @xmath180 . recall that the set of colours @xmath174 correspond to the distinguished generators @xmath181 of the connection group of @xmath5 , and that by @xmath132 we mean the edge between vertices @xmath120 and @xmath121 of colour @xmath4 .
since @xmath125 is a connected graph , without loss of generality , we can suppose that there is at least one edge joining @xmath182 and @xmath183 and another joining @xmath183 and @xmath184 .
let @xmath185 be the colours of these edges , respectively .
that is , without loss of generality we may assume that @xmath186 and @xmath187 are edges of @xmath125 . by lemma [ 2factors4vertices ] , we must have that @xmath188 , as otherwise @xmath125 would have to have at least 4 vertices .
this implies that the only edges of @xmath125 are either @xmath186 and @xmath189 , @xmath186 and @xmath190 or @xmath186 , @xmath189 and @xmath190 , with @xmath191 .
( see figure [ i - transitive3 ] ) .
-maniplexes with edges of colours @xmath192 , @xmath8 , and @xmath193 , with @xmath191.,width=453 ] an easy computation now shows that there are @xmath178 possible different symmetry type graphs of 3-orbit maniplexes of rank @xmath11 . given a 3-orbit @xmath110-maniplex @xmath5 with symmetry type graph having exactly two edges @xmath194 and @xmath195 of colours @xmath8 and @xmath193 , respectively , for some @xmath196 , we shall say that @xmath5 is in class @xmath197 . if , on the other hand , the symmetry type graph of @xmath5 has one edge of colour @xmath8 and parallel edges of colours @xmath192 and @xmath193 , for some @xmath198 , then we say that @xmath5 is in class @xmath199 . from figure [ i - transitive3 ]
we observe that a maniplex in class @xmath200 is @xmath4-face - transitive whenever @xmath201 , while a maniplex in class @xmath199 if @xmath4-face - transitive for every @xmath202 .
a @xmath203-orbit maniplex is @xmath8-face - transitive if and only if it does not belong to any of the classes @xmath199 , @xmath197 or @xmath204 .
[ no3-orbitfully ] there are no fully - transitive 3-orbit maniplexes . using proposition [ stgoffaces ]
, we get some information about the number of flag orbits that the @xmath8-faces have : a @xmath203-orbit maniplex in class @xmath199 or @xmath197 has reflexible @xmath8-faces . if @xmath5 is a @xmath203-orbit maniplex , then the orbits of the @xmath8-faces correspond to the connected components of @xmath205 .
assuming that @xmath5 is in class @xmath199 or @xmath197 , the graph @xmath205 has two connected components ; an isolated vertex , and two vertices that are connected by a @xmath21-edge ( and a @xmath19-edge , if @xmath5 is in class @xmath197 .
then by proposition [ stgoffaces ] , the @xmath8-faces that correspond to the isolated vertex are reflexible ( that is , 1-orbit ) , and the edge with label @xmath193 forces an identification between the two vertices of the second component , so the @xmath8-faces in that component are also reflexible .
it does not take long to realise that counting the number of symmetry type graphs with @xmath206 vertices , and perhaps classifying them in a similar fashion as was done for 2 and 3 vertices , becomes considerably more difficult . in this section
, we shall analyse symmetry type graphs with 4 vertices and determine how far a 4-orbit maniplex can be from being fully - transitive . the following lemma is a consequence of the fact that by taking away the @xmath4-edges of a symmetry type graph @xmath207 ) , the resulting @xmath145 can not have too many components . [ 4orbmi - faceorb ] let @xmath5 be a 4-orbit @xmath110-maniplex and let @xmath208 .
then @xmath5 has one , two or three orbits of @xmath4-faces .
if an @xmath110-maniplex @xmath5 is not fully - transitive , there exists at least one @xmath208 such that @xmath145 is disconnected . we shall divide the analysis of the types in three parts : when @xmath209 has three connected components ( two of them of one vertex and one with two vertices ) , when @xmath145 has a connected component with one vertex and another connected component with three vertices , and finally when @xmath209 has two connected components with two vertices each .
before we start the case analysis , we let @xmath210 be the vertices of @xmath125 .
suppose that @xmath209 has three connected components with @xmath183 and @xmath184 in the same component .
without loss of generality we may assume that @xmath125 has edges @xmath211 and @xmath212 .
let @xmath213 be the colour of an edge between @xmath183 and @xmath184 . since there is no edge of @xmath125 between @xmath182 and @xmath214 , lemma [ 2factors4vertices ] implies that there are at most two such possible @xmath0 , namely @xmath215 and @xmath216 .
if @xmath217 , @xmath125 can have either both edges or exactly one of them , while if @xmath218 there is one possible edge ( see figure [ 3components4 ] ) .
-maniplex @xmath5 with four orbits on its flags , and three orbits on its @xmath4-faces.,width=377 ] let us now assume that @xmath209 has two connected components , one consisting of the vertex @xmath182 and the other one containing vertices @xmath219 and @xmath214 .
this means that the @xmath4-edge incident to @xmath182 is the unique edge that connects this vertex with the rest of the graph and , without loss of generality , @xmath125 has the edge @xmath211 . as with the previous case ,
lemma [ 2factors4vertices ] implies that an edge between @xmath183 and @xmath184 has colour either @xmath220 or @xmath221 .
first observe that having either @xmath222 or @xmath223 in @xmath125 immediately implies ( by lemma [ 2factors4vertices ] ) that there is no edge between @xmath183 and @xmath214 . now
, if both edges @xmath222 and @xmath223 are in @xmath125 , then an edge between @xmath184 and @xmath214 would have to have colour @xmath4 , contradicting the fact that @xmath145 has two connected components .
hence , there is exactly one edge between @xmath183 and @xmath184 .
it is now straightforward to see that @xmath125 should be as one of the graphs in figure [ 2components4 ] , implying that there are exactly four symmetry type graphs with these conditions for each @xmath224 , but only two symmetry type graph of this kind when @xmath225 or @xmath11 .
-maniplexes with four orbits on its flags , and two orbits on its @xmath4-faces such that one contains three flag orbits and the other contains a single flag orbit.,width=264 ] it is straightforward to see from figure [ 2components4 ] that the next lemma follows .
let @xmath5 be a 4-orbit @xmath110-maniplex with two orbits of @xmath4-faces such that @xmath209 has a connected component consisting of one vertex , and another one consisting of three vertices .
then either @xmath226 or @xmath227 has two connected components , each with two vertices . finally , we turn out our attention to the case where @xmath209 has two connected components , with two vertices each .
suppose that @xmath182 and @xmath183 belong to one component , while @xmath184 and @xmath214 belong to the other .
as the two components must be connected by the edges of colour @xmath4 , we may assume that @xmath228 is an edge of @xmath125 .
if the vertices @xmath183 and @xmath214 have semi - edges of colour @xmath4 , lemma [ 2factors4vertices ] implies that @xmath125 is one of the graphs shown in figure [ 2 - 2components4 ] .
-maniplexs with four orbits on its flags , and two orbits on its @xmath4-faces such that each contains two flag orbits.,width=377 ] on the other hand , if @xmath228 and @xmath229 are both edges of @xmath125 , given @xmath230 , we use again lemma [ 2factors4vertices ] to see that @xmath186 is an edge of @xmath125 if and only if @xmath231 is also an edge of @xmath125 .
by contrast , @xmath125 can have either two edges of colour @xmath232 ( each joining the vertices of each connected component of @xmath209 ) , four semi - edges or an edge and two semi - edges of colour @xmath232 .
hence , if @xmath217 , for each @xmath233 there are ten symmetry type graph with semi - edges of colours in @xmath234 and edges of colours not in @xmath234 , as shown in figures [ 3(n-2)_2compa ] and [ 3(n-2)_2compb ] , while for @xmath235 there are six such graphs ( shown in figure [ 3(n-2)_2compb ] ) . on the other hand
if @xmath236 , for each @xmath233 there are two graphs as in figure [ 3(n-2)_2compa ] and one as in figure [ 3(n-2)_2compb ] , while for @xmath235 , there is only one of the graphs in figure [ 3(n-2)_2compb ] .
-maniplexes with four orbits on its flags , and two orbits on its @xmath4-faces such that each contains two flag orbits.,width=264 ] -maniplexes with four orbits on its flags , and two orbits on its @xmath4-faces such that each contains two flag orbits.,width=453 ] we summarize our analysis of the transitivity of 4-orbit maniplexes below .
let @xmath5 be a 4-orbit maniplex .
then , one of the following holds . 1 .
@xmath5 is fully - transitive .
there exists @xmath208 such that @xmath5 is @xmath8-face - transitive for all @xmath237 .
there exist @xmath238 , @xmath239 , such that @xmath5 is @xmath8-face - transitive for all @xmath240 .
there exists @xmath208 such that @xmath5 is @xmath8-face - transitive for all @xmath241 .
every 1-maniplex is reflexible and hence fully - transitive .
fully - transitive 2-maniplexes correspond to fully - transitive maps .
it is well - known ( and easy to see from the symmetry type graph ) that if a map is edge - transitive , then it should have one , two or four orbits .
moreover , a fully - transitive map should be regular , a two - orbit map in class 2 , @xmath242 , @xmath243 or @xmath244 , or a four - orbit map in class @xmath245 or @xmath246 ( see , for example , @xcite ) .
when considering fully - transitive @xmath2-maniplexes , @xmath247 , the analysis becomes considerably more complicated . in @xcite hubard
shows that there are @xmath248 classes of fully - transitive two - orbit @xmath2-maniplexes . by theorem [ no3-orbitfully ] ,
there are no 3-orbit fully - transitive @xmath2-maniplexes .
we note that there are 20 symmetry type graphs of 4-orbit 3-maniplexes that are fully transitive ( see figure [ 4orbitfully ] ) .
the following theorem shall be of great use to show that a fully - transitive 3-maniplex must have an even number of flag orbits unless it is reflexible .
let @xmath5 be a fully - transitive 3-maniplex and let @xmath125 be its symmetry type graph .
then either @xmath5 is reflexible or @xmath125 has an even number of vertices . on the contrary
suppose that @xmath125 has an odd number of vertices , different than 1 .
whenever @xmath249 , the connected components of the @xmath168 2-factor of a symmetry type graph are as in figure [ 4cyclequotient ] . hence , there is a connected component of the @xmath250 2-factor of @xmath125 with exactly one vertex @xmath121 ( and , hence , semi - edges of colours 0 and 2 ) .
the connectivity of @xmath125 implies that there is a vertex @xmath182 adjacent to @xmath121 in @xmath125 .
if @xmath182 is the only neighbour of @xmath121 , then @xmath125 has the edges @xmath251 and @xmath252 as otherwise @xmath5 is not fully - transitive .
since the connected components of the @xmath253 2-factor of @xmath125 are as in figure [ 4cyclequotient ] , @xmath182 has a 0 coloured semi - edge .
because @xmath125 has more than two vertices , the edge of @xmath182 of colour 2 joins @xmath182 to another vertex , say @xmath120 . but removing the edge @xmath254 disconnects the graph contradicting the fact that @xmath5 is 2-face - transitive . on the other hand ,
if @xmath121 has more than one neighbour it has exactly two , say @xmath182 and @xmath120 and @xmath125 has the two edges @xmath251 and @xmath255 .
this implies that the connected component of the @xmath256 2-factor containing @xmath121 has four vertices : @xmath257 and @xmath183 .
( therefore @xmath258 and @xmath259 are edges of @xmath125 . ) using the @xmath253 2-factor one sees that @xmath120 has a semi - edge of colour 0 .
now , if @xmath260 is an edge of @xmath125 , then the vertices @xmath261 and @xmath120 are joined to the rest of @xmath125 by the edges of colour 2 , implying that removing them shall disconnect @xmath125 ( there exists at least another vertex in @xmath125 as it has an odd number of vertices ) , which is again a contradiction .
on the other hand , if @xmath182 ( or @xmath183 ) has an edge of colour 0 to a vertex @xmath184 , then by lemma [ 2factors4vertices ] @xmath183 ( or @xmath182 ) has a 0-edge to a vertex @xmath214 .
again , if @xmath262 is an edge of @xmath125 , since the number of vertices of the graph is odd , removing the edges of colour 2 will leave only the vertices @xmath263 in one component , which is a contradiction .
proceeding now by induction on the number of vertices one can conclude that @xmath125 can not have an odd number of vertices
it is well - known among polytopists that the automorphism group of a regular @xmath2-polytope can be generated by @xmath2 involutions .
in fact , given a base flag @xmath264 , the distinguished generators of @xmath105 with respect to @xmath24 are involutions @xmath265 such that @xmath266 .
generators for the automorphism group of a two - orbit @xmath2-polytope can also be given in terms of a base flag ( see @xcite ) . in this section
we give a set of distinguished generators ( with respect to some base flag ) for the automorphism group of a @xmath0-orbit @xmath110-maniplex in terms of the symmetry type graph @xmath125 , provided that @xmath125 has a hamiltonian path .
given two walks @xmath267 and @xmath268 along the edges and semi - edges of @xmath269 such that the final vertex of @xmath267 is the starting vertex of @xmath268 , we define the sequence @xmath270 as the walk that traces all the edges of @xmath267 and then all the edges of @xmath268 in the same order ; the inverse of @xmath267 , denoted by @xmath271 , is the walk which has the final vertex of @xmath267 as its starting vertex , and traces all the edges of @xmath267 in reversed order . since each of the elements of @xmath272 associated to the edges of @xmath269 is its own inverse , we shall forbid walks that trace the same edge two times consecutively ( or just remove the edge from such walk , shortening its length by two ) .
given a set of walks in @xmath269 , we say that a subset @xmath273 is _ a generating set of @xmath274 _ if each @xmath275 can be expressed as a sequence of elements of @xmath276 and their inverses .
now , let @xmath274 be the set of closed walks along the edges and semi - edges of @xmath269 starting at a distinguished vertex @xmath277 .
recall that the walks along the edges and semi - edges of @xmath269 correspond to permutations of the flags of @xmath58 ; moreover , each closed walk of @xmath274 corresponds to an automorphism of @xmath58 .
thus , by finding a generating set of @xmath274 , we will find a set of automorphisms of @xmath58 that generates @xmath278 .
( however , the converse is not true , as an automorphism of @xmath5 may be described in more than one way as a closed walk of @xmath125 . )
given @xmath269 , we may easily find such generating set .
the construction goes as follows : let @xmath58 be a @xmath0-orbit maniplex of rank @xmath11 such that @xmath279 is a walk of minimal length that visits all the vertices of @xmath125 .
the sets of vertices and edges ( and semi - edges ) of @xmath269 will be denoted by @xmath112 and @xmath280 , respectively .
the set of edges visited by @xmath281 will be denoted by @xmath282 . in this section , the edges joining two vertices @xmath283 and @xmath284
will be denoted by @xmath285 , @xmath286 , @xmath287, ... ,@xmath288 ; if @xmath289 then @xmath290 .
( note that in order to not start carrying many subindices , we modify the notation of the edges of @xmath125 that we had used throughout the paper . if one wants to be consistent with the notation of the edges used in the previous sections , one would have to say that the edges between @xmath291 and @xmath292 are @xmath293 , @xmath294 ) .
similarly , we denote all semi - edges incident to a vertex @xmath283 by @xmath295 , @xmath296 , @xmath297, ... ,@xmath298 . for the sake of simplicity
, @xmath285 will be just called @xmath299 .
let @xmath274 be the set of all closed walks in @xmath269 with @xmath300 as its starting vertex .
we shall now construct @xmath301 , a generating set of @xmath274 .
for each edge @xmath302 we shall define the walk @xmath303 that is , we walk from @xmath277 to @xmath291 in @xmath282 , and then we take the edge @xmath304 , and then we walk back from @xmath292 to @xmath277 in @xmath282
. let @xmath305 be the set of all such walks . for each semi -
edge @xmath306 we shall define the walk @xmath307 .
that is , we walk from @xmath277 to @xmath291 in @xmath282 , and then we take the semi - edge @xmath298 , and then we walk back from @xmath291 to @xmath277 in @xmath282 .
let @xmath308 be the set of all such walks .
we define @xmath309 .
[ generatingwalks ] with the notation from above , @xmath310 is a generating set for @xmath274 .
we shall prove that any @xmath275 can be expressed as a sequence of elements of @xmath310 and their inverses .
let @xmath275 be a closed walk among the edges and semi - edges of @xmath269 starting at @xmath277 .
from now on , semi - edges will be referred to simply as `` edges '' .
we shall proceed by induction over @xmath2 , the number of edges in @xmath311 visited by @xmath43 . if @xmath43 visits only one edge in @xmath311 , then @xmath312 or @xmath313 .
let us suppose that , if a closed walk among the edges of @xmath269 visits @xmath314 different edges in @xmath311 , with @xmath315 , then it can be expressed as a sequence of elements of @xmath310 and their inverses .
let @xmath275 be a walk that visits exactly @xmath2 edges in @xmath311 .
let @xmath316 be the last edge of @xmath311 visited by @xmath43 .
without loss of generality we may assume that the vertex @xmath317 was visited after @xmath318 , so let @xmath319 be the edge that @xmath43 visits just before @xmath320 ( note that @xmath319 may or may not be in @xmath282 ) .
let @xmath321 be the closed walk that traces the same edges ( in the same order ) as @xmath43 until reaching @xmath322 and then traces the edges @xmath323 , @xmath324 , ... ,
@xmath325 , and let @xmath326 be the closed walk that traces the edges @xmath327 and then traces @xmath320 and continues the way @xmath43 does to return to @xmath277 .
it is clear that @xmath267 visits exactly @xmath11 edges in @xmath311 and that @xmath268 visits only one . by inductive hypothesis
both @xmath267 and @xmath268 can be expressed as a sequence of elements of @xmath310 , and therefore so does @xmath43 since @xmath328 .
let @xmath24 be a base flag of @xmath5 that projects to the initial vertex of a walk that contains all vertices of @xmath125 of a symmetry type graph .
following the notation of @xcite , given @xmath329 such that @xmath330 is in the same orbit as @xmath24 ( that is , @xmath331 ) , we denote by @xmath332 the automorphism taking @xmath24 to @xmath330 .
moreover , if @xmath333 for some @xmath334 , then we may also denote @xmath332 by @xmath335 .
the following theorem gives distinguished generators ( with respect to some base flag ) of the automorphism group of a maniplex @xmath5 in terms of a distinguished walk of @xmath125 , that travels through all the vertices of @xmath125 .
its proof is a consequence of the previous lemma .
[ auto ] let @xmath5 be a @xmath0-orbit @xmath2-maniplex and let @xmath125 its symmetry type graph .
suppose that @xmath336 is a distinguished walk that visits every vertex of @xmath125 , with the edge @xmath337 having colour @xmath338 , for each @xmath339 .
let @xmath340 be such that @xmath291 has a semi - edge of colour @xmath341 if and only if @xmath342 .
let @xmath343 be the set of colours of the edges between the vertices @xmath291 and @xmath292 ( with @xmath344 ) that are not in the distinguished walk and let @xmath345 be a base flag of @xmath5 such that @xmath24 projects to @xmath182 in @xmath125 .
then , the automorphism group of @xmath5 is generated by the union of the sets @xmath346 and @xmath347 we note that , in general , a set of generators of @xmath105 obtained from theorem [ auto ] can be reduced since there might be more than one element of @xmath310 representing the same automorphism . for example , the closed walk @xmath43 through an edge of colour @xmath177 , then a @xmath9-semi - edge and finally a 2-edge corresponds to the element @xmath348 of @xmath74 .
hence , the group generator induced by the walk @xmath43 is the same as that induced by the closed walk consisting only of the semi - edge of colour @xmath9 .
the following two corollaries give a set of generators for 2- and 3-orbit polytopes , respectively , in a given class .
the notation follows that of theorem [ auto ] , where if the indices of some @xmath79 do not fit into the parameters of the set , we understand that such automorphism is the identity . *
@xcite * let @xmath5 be a 2-orbit @xmath110-maniplex in class @xmath176 , for some @xmath349 and let @xmath350 .
then @xmath351 is a generating set for @xmath105 .
let @xmath5 be a 3-orbit @xmath110-maniplex . 1 . if @xmath5 is in class @xmath352 , for some @xmath353 , then @xmath354 is a generating set for @xmath105 .
if @xmath5 is in class @xmath355 , for some @xmath356 , then @xmath357 is a generating set for @xmath105 .
a maniplex @xmath5 is said to be _
orientable _ if its flag graph @xmath109 is a bipartite graph .
since a subgraph of a bipartite graph is also bipartite , all the sections of an orientable maniplex are orientable maniplexes themselves . an _ orientation _ of an orientable maniplex
is a colouring of the parts of @xmath109 , with exactly two colours , say black and white .
oriented maniplex _ is an orientable maniplex with a given orientation .
note that any oriented maniplex @xmath5 has an enantiomorphic maniplex ( or mirror image ) @xmath358 .
one can think of the enantiomorphic form of an oriented maniplex simply as the orientable maniplex with the opposite orientation .
if the connection groups @xmath74 of @xmath5 is generated by @xmath181 , for each @xmath359 let us define the element @xmath360 .
then , @xmath361 , for @xmath362 .
the subgroup @xmath363 of @xmath74 generated by @xmath364 is called _ even connection group of @xmath5_. note that @xmath363 has index at most two in @xmath74 .
in fact @xmath365 .
it should be clear then that any maniplex and its enantiomorphic form are in fact isomorphic as maniplexes . an _ oriented flag di - graph _
@xmath366 of an oriented maniplex @xmath5 is constructed in the following way .
the vertex set of @xmath366 consists of one of the parts of the bipartition of @xmath109 .
that is , the black ( or white ) vertices of the flag graph of @xmath5 .
the darts of @xmath366 will be the 2-arcs of @xmath109 of colours @xmath367 , for each @xmath368 .
we then identify two darts to obtain an edge if they have the same vertices , but go in opposite directions .
note that for @xmath369 and each flag @xmath24 of @xmath5 , the 2-arc starting at @xmath24 and with edges coloured @xmath11 and @xmath4 has the same end vertex than the 2-arc starting at @xmath24 and with edges coloured @xmath4 and @xmath11 .
hence , all the darts corresponding to 2-arcs of colours @xmath370 and @xmath4 , with @xmath362 will have both directions in @xmath366 giving us , at each vertex , @xmath371 different edges . on the other hand , the 2-arcs on edges of two colours @xmath372 will in general
be directed darts of @xmath366 .
an example of an oriented flag di - graph is shown in firgure [ orientedgraphflag ] .
we note that the oriented flag di - graph of @xmath358 can be obtained from @xmath366 by reversing the directions of the @xmath373 darts .
note that the 2-arcs of colours @xmath374 correspond to the generators @xmath375 of @xmath363 .
in fact , as @xmath363 consists precisely of the even words of @xmath74 , a maniplex is orientable if and only if the index of @xmath363 in @xmath74 is exactly two .
we can then colour the edges and darts of @xmath366 with the elements @xmath375 .
the fact that @xmath361 for every @xmath376 indeed implies that the edges of @xmath366 are labelled by these first @xmath371 elements , while the darts are labelled by @xmath377 .
we can see now that for each @xmath359 , the @xmath4-faces of @xmath5 are in correspondence with the connected components of the subgraph of @xmath366 with edges of colours @xmath378 . to identify the facets of @xmath5 as subgraphs of @xmath366
, we first consider some oriented paths on the edges of @xmath366 .
we shall say that an oriented path on the edges of @xmath366 is _ facet - admissible _ if no two darts of colour @xmath377 are consecutive on the path .
then , two vertices of @xmath366 are in the same facet of @xmath5 if there exists a facet admissible oriented path from one of the vertices to the other .
for the remainder of this section , by a maniplex we shall mean an oriented maniplex , with one part of the flags coloured with black and the other one in white .
an _ orientation preserving automorphism _ of an ( oriented ) maniplex @xmath5 is an automorphism of @xmath5 that sends black flags to black flags and white flags to white flags .
an _ orientation reversing automorphism _ is an automorphism that interchanges black and white flags .
a _ reflection _ is an orientation reversing involutory automorphism .
the group of orientation preserving automorphisms of @xmath5 shall be denoted by @xmath379 . the orientation preserving automorphism @xmath379 of a maniplex @xmath5
is a subgroup of index at most two in @xmath105 .
in fact , the index is exactly two if and only if @xmath105 contains an orientation reversing automorphism .
note that in this case , there exists an orientation reversing automorphism that sends @xmath5 to its enantiomorphic form @xmath358 .
pisanski @xcite defines a maniplex to be _ chiral - a - la - conway _ if @xmath380 .
if a maniplex @xmath5 is chiral - a - la - conway , then its enantiomorphic maniplex @xmath358 is isomorphic to @xmath5 , but there is no automorphism of the maniplex sending one to the other .
it follows from the definition that @xmath5 is chiral - a - la - conway if and only if the automorphisms of @xmath5 preserve the bipartition of @xmath109 and therefore we have the following proposition . [ oddcycles ]
let @xmath5 be an oriented maniplex and let @xmath125 its symmetry type graph .
then , @xmath5 is chiral - a - la - conway if and only if @xmath125 has no odd cycles .
similarly as before , the orientation preserving automorphisms of a maniplex @xmath5 correspond to colour preserving automorphism of the bipartite graph @xmath109 that preserves the two parts . but these correspond to colour preserving automorphisms of the di - graph @xmath366 , implying that @xmath381 .
note that the action of @xmath379 on the set @xmath382 of all the black flags of @xmath5 is semiregular , and hence , the action on @xmath383 is semiregular on the vertices of @xmath366 .
an oriented maniplex @xmath5 is said to be _ rotary ( or orientably regular ) _ if the action of @xmath379 is regular on @xmath384 .
equivalently , @xmath5 is rotary if the action of @xmath383 is regular on its vertices .
we say that @xmath5 is _
orientably @xmath0-orbit _ if the action of @xmath383 has exactly @xmath0 orbits on the vertices of @xmath366 .
the following lemma is straightforward .
[ orientablekorbit ] let @xmath5 be a chiral - a - la - conway maniplex .
then @xmath125 has no semi - edges and if @xmath5 is an orientably @xmath0-orbit maniplex , then @xmath5 is a @xmath385-orbit maniplex .
we now consider the semiregular action of @xmath379 on the vertices of @xmath366 , and let @xmath386 be the partition of the vertex set of @xmath366 into the orbits with respect to the action of @xmath379 .
( as before , since the action is semiregular , all orbits are of the same size . )
the _ oriented symmetry type di - graph _
@xmath387 of @xmath5 is the quotient colour di - graph with respect to @xmath388 .
similarly as before , if @xmath5 is rotary , then the oriented symmetry type di - graph of @xmath5 consists of one vertex with one loop and @xmath371 semi - edges .
note that for oriented symmetry type di - graphs we shall not identify two darts with the same vertices , but different directions .
if we now turn our attention to oriented symmetry type di - graphs with two vertices , one can see that for each @xmath389 , there is an oriented symmetry type di - graphs with two vertices having semi - edges ( or loops ) of colours @xmath4 at each vertex for every @xmath390 , and having edges ( or both darts ) of colour @xmath8 , for each @xmath391 .
an oriented maniplex with such oriented symmetry type di - graph shall be say to be in class @xmath392 . hence , there are @xmath393 classes of oriented 2-orbit @xmath110-maniplexes .
note that if @xmath5 is a @xmath0-orbit maniplex , then @xmath387 has either @xmath0 or @xmath394 vertices .
the next result follows from proposition [ oddcycles ] and lemma [ orientablekorbit ] .
let @xmath5 be an oriented maniplex .
then , @xmath125 and @xmath387 have the same number of vertices if and only if @xmath125 has a semi - edge or an odd cycle .
it is not difficult to see that if we are to consider for a moment an oriented symmetry type di - graph @xmath395 with an ( undirected ) hamiltonian path , then the construction of section [ gen - autg ] gives us a way to construct a generating set of the closed walks based at the starting vertex of the path ( and lemma [ generatingwalks ] implies that the set actually generates . )
hence , one can find generators for the group of orientation preserving automorphisms of an oriented maniplex , provided that it has an ( undirected ) hamiltonian path . in particular we have the following theorem .
let @xmath5 be an oriented 2-orbit @xmath110-maniplex in class @xmath392 , for some @xmath389
. then 1 .
if @xmath396 , let @xmath350 , then @xmath397 is a generating set for @xmath398 .
2 . if @xmath399 but there exists @xmath350 , @xmath400 , then @xmath401 is a generating set for @xmath398 .
if @xmath402 , then @xmath403 is a generating set for @xmath398 . given an oriented maniplex @xmath5 and its symmetry type graph @xmath125 , we shall say that @xmath387 is the associated oriented symmetry type di - graph of @xmath125 .
hence , given a symmetry type graph @xmath147 one can find its associated oriented symmetry type di - graph @xmath395 by erasing all edges of @xmath147 and replacing them by the @xmath367 paths of @xmath147 .
note that this replacement of the edges may disconnect the new graph .
if that is the case , we take @xmath395 to be one of the connected components . in a similar way
as one can classify maniplexes with small number of flag orbits ( under the action of the automorphism group of the maniplex ) in terms of their symmetry type graph , one can classify oriented maniplexes with small number of flags ( under the action of the orientation preserving automorphism group of the maniplex ) in terms of their oriented symmetry type di - graph .
let @xmath58 be a @xmath404-orbit chiral - a - la - conway @xmath110-maniplex , with @xmath405 .
let @xmath269 be its symmetry type graph and @xmath406 be its oriented symmetry type di - graph . recall that @xmath269 is a graph with 6 vertices and no semi - edges or odd cycles , and that @xmath406 is a di - graph with 3 vertices .
let @xmath407 be the vertex set of @xmath269 .
we may label the vertices of @xmath125 in such a way that the edges @xmath408 , @xmath409 , @xmath410 are coloured with the colour @xmath110 , and that no two vertices of the set @xmath411 are adjacent .
let @xmath412 be the vertex set of @xmath413 .
each @xmath414 corresponds to the vertex @xmath415 , @xmath416 . in what follows , in the same way as in section
[ sec : stg ] , @xmath417 denotes the @xmath0-coloured edge joining the vertices @xmath283 and @xmath284 , @xmath418 , @xmath419 ; and @xmath420 denotes the @xmath421-coloured edge joining the vertices @xmath422 and @xmath423 , @xmath424 and @xmath425 .
since there are no semi - edges in @xmath125 , for each colour @xmath426 there is one edge ( and one semi - edge ) of colour @xmath427 in @xmath413 if and only if the 2-factor of @xmath125 of colours @xmath4 and @xmath110 consists of one 4-cycle and one 2-cycle of alternating colours .
likewise , there are three semi - edges of colour @xmath427 in @xmath413 if and only if the 2-factor of @xmath125 of colours @xmath4 and @xmath110 consist of three 2-cycles .
it is straightforward to see that there are two consecutive edges of colour @xmath427 and @xmath428 @xmath63 , in @xmath413 if and only if the 2-factor of colours @xmath4 and @xmath8 consists of a single 6-cycle .
it follows that if there are two consecutive edges of colour @xmath427 and @xmath429 in @xmath413 , then @xmath430 .
notice that the possible 2-factors of colour @xmath110 and @xmath431 in @xmath125 are either a single 6-cycle of alternating colours , a 4-cycle along with a 2-cycle , or three separate 2-cycles .
hence , the darts in @xmath413 are arranged in either a 3-cycle , a 2-cycle along with a loop , or three separate loops .
we proceed case by case .
consider the case when there are three loops in @xmath413 .
since oriented symmetry type di - graphs are connected , then without loss of generality @xmath432 and @xmath433 must be edges of @xmath413 .
we may suppose that @xmath432 is the only edge joining @xmath267 and @xmath434 . if there is a third edge in @xmath413 , then it is necessarily @xmath435 .
note that , since the edges coloured by @xmath110 and @xmath431 do not lie on a 6-cycle in @xmath125 , there are no restrictions on the semi - edges of @xmath413 .
thus , there is one oriented symmetry type di - graph for each pair of colours @xmath4 and @xmath221 , with @xmath426 and one for each triple @xmath220 , @xmath4 and @xmath221 , @xmath436 .
therefore , there are @xmath437 oriented symmetry type di - graphs with 3 loops .
consider the case when @xmath413 has only one loop .
we may suppose that the loop is in @xmath438 and the vertices @xmath267 or @xmath434 are joined by darts .
this implies that @xmath439 , @xmath440 and @xmath441 are edges of @xmath125 .
as @xmath413 is connected , there must be an edge joining @xmath434 and @xmath438 of colour @xmath442 necessarily @xmath443 , since the edges @xmath444 , @xmath440 , @xmath445 , @xmath446 , @xmath447 , @xmath448 form a 6-cycle in @xmath125 .
notice that there are no restrictions on the semi - edges of @xmath413 .
hence , there are exactly two oriented symmetry type di - graph with a single loop : one with a single edge of colour @xmath449 between @xmath450 and @xmath451 , and one with two edges of colours @xmath449 and @xmath452 between them . consider the case when the darts in @xmath413 are arranged in a 3-cycle .
it is clear that the 2-factor of @xmath125 of colours @xmath431 and @xmath110 is a single 6-cycle .
therefore , if @xmath453 , the 2-factor of @xmath125 of colours @xmath4 and @xmath110 can not consist of three 2-cycles , as this implies the existence of a 6-cycle of alternating colours @xmath4 and @xmath431 such that @xmath454 .
that is , @xmath413 has one edge ( and one semi - edge ) of colour @xmath427 for each @xmath453 and either one edge and a semi - edge , or three semi - edges for colour @xmath455 note that if @xmath456 , the set @xmath457 has more than three elements and thus all edges of colour @xmath427 in @xmath413 , @xmath453 , must be joining the same pair of vertices .
otherwise , there would be at least two consecutive edges of colours @xmath427 and @xmath429 , with @xmath458 .
figure [ 3orient ] below shows the only four possible oriented symmetry type di - graphs with a 3-cycle of darts and at least two consecutive edges .
two correpond to 4-maniplexes , one to 3-maniplexes and one to 5-maniplexes .
these will be treated as special cases .
we may suppose that @xmath413 has no consecutive edges .
it follows that here are exactly two oriented symmetry type di - graph with a 3-cycle of darts : one with an edge joining the same pair of vertices for each colour @xmath459 and one with three semi - edges of colour @xmath449 and an edge joining the same pair of vertices for each colour @xmath453 .
considering all the cases above , there are @xmath460 oriented symmetry type graphs with three vertices for oriented maniplexes of rank @xmath461 ; @xmath462 for oriented maniplexes of rank 3 ; @xmath463 for oriented maniplexes of rank 4 ; and @xmath464 for oriented maniplexes of rank 5 .
this work was done with the support of programa de apoyo a proyectos de investigacin e innovacin tecnolgica ( papiit ) de la unam , ib101412 _ grupos y grficas asociados a politopos abstractos _ `` .
the second author was partially supported by slovenian research agency ( arrs ) and the third author was partially supported by conacyt under project 166951 and by the program para las mujeres en la ciencia loreal - unesco - amc 2012 '' | a @xmath0-orbit maniplex is one that has @xmath0 orbits of flags under the action of its automorphism group . in this paper
we extend the notion of symmetry type graphs of maps to that of maniplexes and polytopes and make use of them to study @xmath0-orbit maniplexes , as well as fully - transitive 3-maniplexes .
in particular , we show that there are no fully - transtive @xmath0-orbit 3-mainplexes with @xmath1 an odd number , we classify 3-orbit mainplexes and determine all face transitivities for 3- and 4-orbit maniplexes .
moreover , we give generators of the automorphism group of a polytope or a maniplex , given its symmetry type graph . finally , we extend these notions to oriented polytopes , in particular we classify oriented 2-orbit maniplexes and give generators for their orientation preserving automorphism group . |
six years after the financial crisis of 2007 - 2008 , millions of households worldwide are still struggling to recover from the aftermath of those traumatic events .
the majority of losses are indirect , such as people losing homes or jobs , and for the majority , income levels have dropped substantially . for the economy as a whole , and for households and for public budgets , the miseries of the market meltdown of 2007 - 2008 are not yet over.as a consequence , a consensus for the need for new financial regulation is emerging @xcite .
future financial regulation should be designed to mitigate risks within the financial system as a whole , and should specifically address the issue of systemic risk ( sr ) .
sr is the risk that the financial system as a whole , or a large fraction thereof , can no longer perform its function as a credit provider , and as a result collapses . in a narrow sense , it is the notion of contagion or impact from the failure of a financial institution or group of institutions on the financial system and the wider economy @xcite .
generally , it emerges through one of two mechanisms , either through interconnectedness or through the synchronization of behavior of agents ( fire sales , margin calls , herding ) .
the latter can be measured by a potential capital shortfall during periods of synchronized behavior where many institutions are simultaneously distressed @xcite .
measures for a potential capital shortfall are closely related to the leverage of financial institutions @xcite .
interconnectedness is a consequence of the network nature of financial claims and liabilities @xcite .
several studies indicate that financial network measures could potentially serve as early warning indicators for crises @xcite .
in addition to constraining the ( potentially harmful ) bilateral exposures of financial institutions , there are essentially two options for future financial regulation to address the problem @xcite : first , financial regulation could attempt to reduce the financial fragility of `` super - spreaders '' or _ systemically important financial institutions _ ( sifis ) , i.e. limiting a potential capital shortfall .
this can achieved by reducing the leverage or increasing the capital requirements for sifis .
`` super - spreaders '' are institutions that are either too big , too connected or otherwise too important to fail .
however , a reduction of leverage simultaneously reduces efficiency and can lead to pro - cyclical effects @xcite .
second , future financial regulation could attempt strengthening the financial system as a whole .
it has been noted that different financial network topologies have different probabilities for systemic collapse @xcite . in this sense
the management of sr is reduced to the technical problem of re - shaping the topology of financial networks @xcite .
the basel committee on banking supervision ( bcbs ) recommendation for future financial regulation for sifis is an example of the first option .
the basel iii framework recognizes sifis and , in particular , global and domestic systemically important banks ( g - sibs or d - sibs ) .
the bcbs recommends increased capital requirements for sifis the so called `` sifi surcharges '' @xcite .
they propose that sr should be measured in terms of the impact that a bank s failure can have on the global financial system and the wider economy , rather than just the risk that a failure could occur .
therefore they understand sr as a global , system - wide , loss - given - default ( lgd ) concept , as opposed to a probability of default ( pd ) concept . instead of using quantitative models to estimate sr
, basel iii proposes an indicator - based approach that includes the size of banks , their interconnectedness , and other quantitative and qualitative aspects of systemic importance .
there is not much literature on the problem of dynamically re - shaping network topology so that networks adapt over time to function optimally in terms of stability and efficiency .
a major problem in nw - based sr management is to provide agents with incentives to re - arrange their local contracts so that global ( system - wide ) sr is reduced .
recently , it has been noted empirically that individual transactions in the interbank market alter the sr in the total financial system in a measurable way @xcite .
this allows an estimation of the marginal sr associated with financial transactions , a fact that has been used to propose a tax on systemically relevant transactions @xcite .
it was demonstrated with an agent - based model ( abm ) that such a tax the _ systemic risk tax _ ( srt ) is leading to a dynamical re - structuring of financial networks , so that overall sr is substantially reduced @xcite . in this paper
we study and compare the consequences of two different options for the regulation of sr with an abm . as an example for the first option we study basel iii with capital surcharges for g - sibs and compare it with an example for the second option the srt that leads to a self - organized re - structuring of financial networks .
a number of abms have been used recently to study interactions between the financial system and the real economy , focusing on destabilizing feedback loops between the two sectors @xcite .
we study the different options for the regulation of sr within the framework of the crisis macro - financial model . in this abm , we implement both the basel iii indicator - based measurement approach , and the increased capital requirements for g - sibs .
we compare both to an implementation of the srt developed in @xcite .
we conduct three `` computer experiments '' with the different regulation schemes .
first , we investigate which of the two options to regulate sr is superior .
second , we study the effect of increased capital requirements , the `` surcharges '' , on g - sibs and the real economy .
third , we clarify to what extend the basel iii indicator - based measurement approach really quantifies sr , as indented by the bcbs .
the basel iii indicator - based measurement approach consists of five broad categories : size , interconnectedness , lack of readily available substitutes or financial institution infrastructure , global ( cross - jurisdictional ) activity and `` complexity '' .
as shown in [ indicator ] , the measure gives equal weight to each of the five categories .
each category may again contain individual indicators , which are equally weighted within the category .
p5 cm p7 cm p3 cm category ( and weighting ) & individual indicator & indicator weighting + cross - jurisdictional activity ( 20% ) & cross - jurisdictional claims & 10% + & cross - jurisdictional liabilities & 10% + size ( 20% ) & total exposures as defined for use in the basel iii leverage ratio & 20% + interconnectedness ( 20% ) & intra - financial system assets & 6.67% + & intra - financial system liabilities & 6.67% + & securities outstanding & 6.67% + substitutability / financial institution infrastructure ( 20% ) & assets under custody & 6.67% + & payments activity & 6.67% + & underwritten transactions in debt and equity markets & 6.67% + complexity ( 20% ) & notional amount of over - the - counter ( otc ) derivatives & 6.67% + & level 3 assets & 6.67% + & trading and available - for - sale securities & 6.67% here below we describe each of the categories in more detail . this indicator captures the global `` footprint '' of banks . the motivation is to reflect the coordination difficulty associated with the resolution of international spillover effects .
it measures the bank s cross - jurisdictional activity relative to other banks activities .
here it differentiates between [ ( i ) ] cross - jurisdictional claims , and cross - jurisdictional liabilities .
this indicator reflects the idea that size matters .
larger banks are more difficult to replace and the failure of a large bank is more likely to damage confidence in the system . the indicator is a measure of the normalized total exposure used in basel iii leverage ratio .
a bank s systemic impact is likely to be related to its interconnectedness to other institutions via a network of contractual obligations .
the indicator differentiates between [ ( i ) ] in - degree on the network of contracts out - degree on the network of contracts , and outstanding securities .
the indicator for this category captures the increased difficulty in replacing banks that provide unique services or infrastructure .
the indicator differentiates between [ ( i ) ] assets under custody payment activity , and underwritten transactions in debt and equity markets .
banks are complex in terms of business structure and operational `` complexity '' . the costs for resolving a complex bank is considered to be greater .
this is reflected in the indicator as [ ( i ) ] notional amount of over - the - counter derivatives level 3 assets , and trading and available - for - sale securities .
the score @xmath0 of the basel iii indicator - based measurement approach for each bank @xmath1 and each indicator @xmath2 , e.g. cross - jurisdictional claims , is calculated as the fraction of the individual banks with respect to all @xmath3 banks and then weighted by the indicator weight ( @xmath4 ) .
the score is given in basis points ( factor @xmath5 ) @xmath6 where @xmath7 is the set of indicators @xmath2 and @xmath4 the weights from [ indicator ] .
p3 cm p3 cm p3 cm p6 cm bucket & score range & bucket thresholds & higher loss absorbency requirement ( common equity as a percentage of risk - weighted assets ) + 5 & d - e & 530 - 629 & 3.50% + 4 & c - d & 430 - 529 & 2.50% + 3 & b - c & 330 - 429 & 2.00% + 2 & a - b & 230 - 329 & 1.50% + 1 & cutoff point - a & 130 - 229 & 1.00% in the basel iii `` bucketing approach '' , based on the scores from [ score ] , banks are divided into four equally sized classes ( buckets ) of systemic importance , seen here in [ buckets ] .
the cutoff score and bucket thresholds have been calibrated by the bcbs in such a way that the magnitude of the higher loss absorbency requirements for the highest populated bucket is 2.5% of risk - weighted assets , with an initially empty bucket of 3.5% of risk - weighted assets .
the loss absorbency requirements for the lowest bucket is 1% of risk - weighted assets .
the loss absorbency requirement is to be met with common equity @xcite .
bucket five will initially be empty .
as soon as the bucket becomes populated , a new bucket will be added in such a way that it is equal in size ( scores ) to each of the other populated buckets and the minimum higher loss absorbency requirement is increased by 1% of risk - weighted assets .
we use an abm , linking the financial and the real economy .
the model consists of banks , firms and households .
one pillar of the model is a well - studied macroeconomic model @xcite , the second pillar is an implementation of an interbank market . in particular , we extend the model used in @xcite with an implementation of the basel iii indicator - based measurement approach and capital surcharges for sibs .
for a comprehensive description of the model , see @xcite .
the agents in the model interact on four different markets .
[ ( i ) ] firms and banks interact on the credit market , generating flows of loan ( re)payments .
banks interact with other banks on the interbank market , generating flows of interbank loan ( re)payments .
households and firms interact on the labour market , generating flows wage payments .
households and firms interact on the consumption - goods market , generating flows of goods .
banks hold all of firms and households ?
cash as deposits .
households are randomly assigned as owners of firms and banks ( share - holders ) .
agents repeat the following sequence of decisions at each time step : firms define labour and capital demand , banks rise liquidity for loans , firms allocate capital for production ( labour ) , households receive wages , decide on consumption and savings , firms and banks pay dividends , firms with negative cash go bankrupt , banks and firms repay loans , illiquid banks try to rise liquidity , and if unsuccessful , go bankrupt . households which own
firms or banks use dividends as income , all other households use wages .
banks and firms pay @xmath8 of their profits as dividends .
the agents are described in more detail below .
there are @xmath9 households in the model .
households can either be workers or investors that own firms or banks .
each household @xmath1 has a personal account @xmath10 at one of the @xmath3 banks , where index @xmath1 represents the household and index @xmath11 the bank .
workers apply for jobs at the @xmath12 different firms .
if hired , they receive a fixed wage @xmath13 per time step , and supply a fixed labour productivity @xmath14 .
a household spends a fixed percentage @xmath15 of its current account on the consumption goods market .
they compare prices of goods from @xmath16 randomly chosen firms and buy the cheapest .
there are @xmath12 firms in the model .
they produce perfectly substitutable goods . at each time step firm
@xmath17 computes its own expected demand @xmath18 and price @xmath19 .
the estimation is based on a rule that takes into account both excess demand / supply and the deviation of the price @xmath20 from the average price in the previous time step @xcite .
each firm computes the required labour to meet the expected demand . if the wages for the respective workforce exceed the firm s current liquidity , the firm applies for a loan .
firms approach @xmath21 randomly chosen banks and choose the loan with the most favorable rate .
if this rate exceeds a threshold rate @xmath22 , the firm only asks for @xmath23 percent of the originally desired loan volume . based on the outcome of this loan request , firms re - evaluate the necessary workforce , and hire or fire accordingly .
firms sell the produced goods on the consumption goods market .
firms go bankrupt if they run into negative liquidity .
each of the bankrupted firms debtors ( banks ) incurs a capital loss in proportion to their investment in loans to the company .
firm owners of bankrupted firms are personally liable .
their account is divided by the debtors _ pro rata_. they immediately ( next time step ) start a new company , which initially has zero equity . their initial estimates for demand @xmath18 and price @xmath19
is set to the respective averages in the goods market .
the model involves @xmath3 banks .
banks offer firm loans at rates that take into account the individual specificity of banks ( modeled by a uniformly distributed random variable ) , and the firms creditworthiness .
firms pay a credit risk premium according to their creditworthiness that is modeled by a monotonically increasing function of their financial fragility @xcite .
banks try to grant these requests for firm loans , providing they have enough liquid resources .
if they do not have enough cash , they approach other banks in the interbank market to obtain the necessary amount .
if a bank does not have enough cash and can not raise the total amount requested on the interbank market , it does not pay out the loan .
interbank and firm loans have the same duration .
additional refinancing costs of banks remain with the firms . each time step firms and banks repay @xmath24 percent of their outstanding debt ( principal plus interest ) .
if banks have excess liquidity they offer it on the interbank market for a nominal interest rate .
the interbank relation network is modeled as a fully connected network and banks choose the interbank offer with the most favorable rate .
interbank rates @xmath25 offered by bank @xmath17 to bank @xmath1 take into account the specificity of bank @xmath17 , and the creditworthiness of bank @xmath1 .
if a firm goes bankrupt the respective creditor bank writes off the respective outstanding loans as defaulted credits .
if the bank does not have enough equity capital to cover these losses it defaults . following a bank default an iterative default - event unfolds for all interbank creditors .
this may trigger a cascade of bank defaults . for simplicity s sake
, we assume a recovery rate set to zero for interbank loans . this assumption is reasonable in practice for short term liquidity @xcite .
a cascade of bankruptcies happens within one time step .
after the last bankruptcy is resolved the simulation is stopped . in the abm ,
we implement the size indicator by calculating the total exposures of banks .
total exposure includes all assets of banks excluding cash , i.e. loans to firms and loans to other banks .
interconnectedness is measured in the model by interbank assets ( loans ) and interbank liabilities ( deposits ) of banks . as a measure for substitutability we use the payment activity of banks .
the payment activity is measured by the sum of all outgoing payments by banks . in the model
we do not have cross - jurisdictional activity and banks are not engaged in selling complex financial products including derivatives and level 3 assets .
we therefore set the weights for global ( cross - jurisdictional ) activity and complexity to zero .
banks in the model have to observe loss absorbency ( capital ) requirements according to basel iii .
they are required to hold 4.5% of common equity ( up from 2% in basel ii ) of risk - weighted assets ( rwas ) .
rwas are calculated according to the standardized approach , i.e. with fixed weights for all asset classes . as fixed weights we use 100% for interbank loans and commercial loans .
we define equity capital of banks in the model as common equity . in the model banks
are allocated to the five buckets shown in [ buckets ] based on the scores obtained by [ score ] .
banks have to meet additional loss absorbency requirements as shown in [ buckets ] .
the score is calculated at every time step @xmath26 and capital requirements must be observed before providing new loans .
the srt is implemented in the model as described in @xcite , for details see [ srt ] and @xcite .
the srt is calculated according to [ srteq_simple ] and is imposed on all interbank transactions .
all other transactions are exempted from srt . before entering a desired loan contraction , loan requesting bank @xmath17 obtains quotes for the srt rates from the central bank , for various potential lenders @xmath1 .
bank @xmath17 chooses the interbank offer from bank @xmath1 with the smallest total rate .
the srt is collected into a bailout fund .
below we present the results of three experiments .
the first experiment focuses on the performance of different options for regulation of sr , the second experiment on the consequences of different capital surcharges for g - sibs , and the third experiment on the effects of different weight distributions for the basel iii indicator - based measurement approach . in order to understand the effects of basel iii capital surcharges for g - sibs on the financial system and the real economy , and to compare it with the srt , we conduct an experiment where we consider three cases : [ ( i ) ] financial system regulated with basel ii , financial system regulated with basel iii , with capital surcharges for g - sibs , financial system with the srt . with basel ii we require banks to hold 2% of common equity of their rwas .
rwas are calculated according to the standardized approach .
the effects of the different regulation policies on total losses to banks @xmath27 ( see [ risk_measures ] ) are shown in [ fig : regulation_type](a ) .
clearly , with basel ii ( red ) fat tails in the loss distributions of the banking sector are visible .
basel iii with capital surcharges for g - sibs slightly reduces the losses ( almost not visible ) .
the srt gets almost completely rid of big losses in the system ( green ) .
this reduction in major losses in the financial system is due to the fact that with the srt the possibility of cascading failure is drastically reduced .
this is shown in [ fig : regulation_type](b ) where the distributions of cascade sizes @xmath28 ( see [ risk_measures ] ) for the three modes are compared . with the basel regulation policies we observe cascade sizes of up to @xmath29 banks , while with the srt
the possibility of cascading failure is drastically reduced .
clearly , the total transaction volume @xmath30 ( see [ risk_measures ] ) in the interbank market is not affected by the different regulation policies .
this is seen in [ fig : regulation_type](c ) , where we show the distributions of transaction volumes . in [ fig : regulation_type](d ) we show the _ sr - profile _ of the financial system given by the rank - ordered debtranks @xmath31 of all banks @xcite .
the sr - profile shows the distribution of systemic impact across banks in the financial system .
the bank with the highest systemic impact is to the very left .
obviously , the srt drastically reduces the systemic impact of individual banks and leads to a more homogeneous spreading of sr across banks .
basel iii with capital surcharges for g - sibs sees a slight reduction in the systemic impact of individual banks .
next , we study the effect of different levels of capital surcharges for g - sibs . here
we consider three different settings : [ ( i ) ] capital surcharges for g - sibs as specified in basel iii ( [ buckets ] ) , double capital surcharges for g - sibs , threefold capital surcharges for g - sibs . with larger capital surcharges for g - sibs , we observe a stronger effect of the basel iii regulation policy on the financial system .
clearly , the shape of the distribution of losses @xmath27 is similar ( [ fig : surcharges](a ) ) .
the tail of the distributions is only reduced due to a decrease in efficiency ( transaction volume ) , as is seen in [ fig : surcharges](c ) .
evidently , average losses @xmath27 are reduced at the cost of a loss of efficiency by roughly the same factor .
this means that basel iii must reduce efficiency in order to show any effect in terms of sr , see [ discussion ] . in [ fig : surcharges](d ) we show the sr - profile for different capital surcharges for g - sibs .
clearly , the systemic impact of individual banks is also reduced at the cost of a loss of efficiency by roughly the same factor . in [ fig : surcharges](b ) we show the cascade size of synchronized firm defaults . here
we see an increase in cascading failure of firms with increasing capital surcharges for g - sibs .
this means larger capital surcharges for g - sibs can have pro - cyclical side effects , see [ discussion ] . finally , we study the effects of different weight distributions for the basel iii indicator - based measurement approach . in particular , we are interested in whether it is more effective to have capital surcharges for `` super - spreaders '' or for `` super - vulnerable '' financial institutions .
a vulnerable financial institution in this context is an institution that is particularly exposed to failures in the financial system , i.e. has large credit or counterparty risk ( cr ) .
holding assets is generally associated with the risk of losing the value of an investment . whereas a liability is an obligation towards a counterparty that can have an impact if not fulfilled . to define an indicator that reflects sr and identifies `` super - spreaders '' we set the weight on intra - financial system liabilities to 100% . for an indicator that reflects
cr of interbank assets and identifies vulnerable financial institutions of we set the weight on intra - financial system assets to 100% .
weights on all other individual indicators are set to zero .
specifically , we consider three different weight distributions .
[ ( i ) ] weights as specified in basel iii ( [ indicator ] ) , indicator weight on interbank liabilities set to 100% , indicator weight on interbank assets set to 100% . to illustrate the effects of different weight distributions
, we once again use larger capital surcharges for g - sibs , as specified in basel iii . to allow a comparison between the different weight distributions , we set the level of capital surcharges for each weight distribution to result in a similar level of efficiency ( credit volume ) .
specifically , for the weight distribution as specified in basel iii ( [ indicator ] ) we multiply the capital surcharges for g - sibs as specified in basel iii ( [ buckets ] ) by a factor @xmath32 , for indicator weight on interbank liabilities by @xmath33 and for indicator weight on interbank assets by @xmath34 . in [ fig : weights](b ) we show that efficiency , as measured by transaction volume in the interbank market , is indeed similar for all weight distributions . in [ fig : weights](a ) we show the losses to banks @xmath27 for the different weight distributions .
clearly , imposing capital surcharges on `` super - spreaders '' ( yellow ) does indeed reduce the tail of the distribution , but does not , however , get completely rid of large losses in the system ( yellow ) .
in contrast , imposing capital surcharges on `` super - vulnerable '' financial institutions ( red ) shifts the mode of the loss distribution without reducing the tail , thus making medium losses more likely without getting rid of large losses . to illustrate the effect of the different weight distributions , we show the risk profiles for sr and cr .
risk profiles for sr show the systemic impact as given by the debtrank @xmath35 ( banks are ordered by @xmath35 , the most systemically important being to the very left ) .
risk profiles for cr show the distribution of _ average vulnerability _ , i.e. a measure for cr in a financial network , for details see [ vulnerability_section ] .
the corresponding sr- and cr - profiles are shown for the different weight distributions in [ fig : weights](c ) and [ fig : weights](d ) .
imposing capital surcharges on `` super - spreaders '' leads to a more homogeneous spreading of sr across all agents , as shown in [ fig : weights](c ) ( yellow ) . whereas capital surcharges for `` super - vulnerable '' financial institutions leads to a more homogeneous spreading of cr ( [ fig : weights](d ) ( red ) ) .
interestingly , homogeneous spreading of cr ( sr ) leads to a more unevenly distributed sr ( cr ) , as seen by comparing [ fig : weights](c ) and [ fig : weights](d ) .
we use an abm to study and compare the consequences of the two different options for the regulation of sr .
in particular we compare financial regulation that attempts to reduce the financial fragility of sifis ( basel iii capital surcharges for g - sibs ) with a regulation policy that aims directly at reshaping the topology of financial networks ( srt ) .
sr emerges in the abm through two mechanisms , either through interconnectedness of the financial system or through synchronization of the behavior of agents .
cascading failure of banks in the abm can be explained as follows .
triggered by a default of a firm on a loan , a bank may suffer losses exceeding its absorbance capacity .
the bank fails and due to its exposure on the interbank market , other banks may fail as well .
cascading failure can be seen in [ fig : regulation_type](b ) .
basel iii capital surcharges for g - sibs work in the abm as follows .
demand for commercial loans from firms depends mainly on the expected demand for goods .
firms approach different banks for loan requests .
if banks have different capital requirements , banks with lower capital requirements may have a higher leverage and provide more loans .
effectively , different capital requirement produce inhomogeneous leverage levels in the banking system . imposing capital surcharges for g - sibs
means that banks with a potentially large impact on others must have sizable capital buffers . with capital surcharges for g - sibs ,
non - important banks can use higher leverage until they become systemically important themselves .
this leads to a more homogenous spreading of sr , albeit that reduction and mitigation of sr is achieved `` indirectly '' .
the srt leads to a self - organized reduction of sr in the following way @xcite : banks looking for credit will try to avoid this tax by looking for credit opportunities that do not increase sr and are thus tax - free . as a result ,
the financial network rearranges itself toward a topology that , in combination with the financial conditions of individual institutions , will lead to a new topology where cascading failures can no longer occur .
the reduction and mitigation of sr is achieved `` directly '' by re - shaping the network . from the experiments we conclude that basel iii capital surcharges for g - sibs reduce and mitigate sr at the cost of a loss of efficiency of the financial system ( [ fig : surcharges](a ) and [ fig : surcharges](c ) ) .
this is because overall leverage is reduced by capital surcharges . on the other hand ,
the srt keeps the efficiency comparable to the unregulated financial system .
however it avoids cascades due to the emergent network structure .
this means potentially much higher capital requirements for `` super - spreaders '' do address sr but at the cost of efficiency .
sr through synchronization emerges in the abm in the following way : similarly to the root cause of cascading failure in the banking system , an initial shock from losses from commercial loans can lead to a synchronization of firm defaults .
if banks are able to absorb initial losses , the losses still result in an implicit increase in leverage . in order to meet capital requirements ( basel iii )
, the bank needs to reduce the volume of commercial loans .
this creates a feedback effect of firms suffering from reduced liquidity , which in turn increases the probability of defaults and closes the cycle .
this feedback effect is inherent to the system and can be re - inforced by basel iii capital surcharges for g - sibs . in the experiments we see that this feedback effect gets more pronounced with larger capital surcharges for g - sibs ( [ fig : surcharges](b ) ) .
here we see an increase in cascading failure of firms with increasing capital surcharges for g - sibs .
this means capital surcharges for g - sibs can have pro - cyclical side effects .
basel iii measures systemic importance with an indicator - based measurement approach .
the indicator approach is based on several individual indicators consisting of banks assets and liabilities . by studying the effect of different weight distributions for the individual indicators
, we find that basel iii capital surcharges for g - sibs are more effective with higher weights on liabilities ( [ fig : weights](a ) ) . this means
quite intuitively that asset - based indicators are more suitable for controlling credit risk and liability based indicators are more suitable for controlling systemic risk .
since the indicators in the basel proposal are predominantly based on assets , the basel iii indicator - based measurement approach captures predominantly credit risk and therefore does not meet it s declared objective . to conclude with a summary at a very high level , the policy implications obtained with the abm are
[ ( i ) ] re - shaping financial networks is more effective and efficient than reducing leverage , capital surcharges for g - sibs ( `` super - spreaders '' ) can reduce sr , but must be larger than specified in basel iii to have an measurable impact , and thus cause a loss of efficiency .
basel iii capital surcharges for g - sibs can have pro - cyclical side effects .
debtrank is a recursive method suggested in @xcite to determine the systemic relevance of nodes in financial networks .
it is a number measuring the fraction of the total economic value in the network that is potentially affected by a node or a set of nodes .
@xmath36 denotes the interbank liability network at any given moment ( loans of bank @xmath1 to bank @xmath17 ) , and @xmath37 is the capital of bank @xmath17 . if bank @xmath17 defaults and can not repay its loans , bank @xmath1 loses the loans @xmath36 . if @xmath1 does not have enough capital available to cover the loss , @xmath1 also defaults .
the impact of bank @xmath17 on bank @xmath1 ( in case of a default of @xmath17 ) is therefore defined as @xmath38 \quad.\ ] ] the value of the impact of bank @xmath17 on its neighbors is @xmath39 .
the impact is measured by the _ economic value _
@xmath40 of bank @xmath17 . for the economic value we use two different proxies .
given the total outstanding interbank exposures of bank @xmath17 , @xmath41 , its economic value is defined as @xmath42 to take into account the impact of nodes at distance two and higher , this has to be computed recursively . if the network @xmath43 contains cycles , the impact can exceed one . to avoid this problem
an alternative was suggested in @xcite , where two state variables , @xmath44 and @xmath45 , are assigned to each node .
@xmath46 is a continuous variable between zero and one ; @xmath47 is a discrete state variable for three possible states , undistressed , distressed , and inactive , @xmath48 .
the initial conditions are @xmath49 , and @xmath50 ( parameter @xmath51 quantifies the initial level of distress : @xmath52 $ ] , with @xmath53 meaning default ) .
the dynamics of @xmath54 is then specified by @xmath55 \quad.\ ] ] the sum extends over these @xmath1 , for which @xmath56 , @xmath57 the debtrank of the set @xmath58 ( set of nodes in distress at time @xmath59 ) , is @xmath60 , and measures the distress in the system , excluding the initial distress .
if @xmath58 is a single node , the debtrank measures its systemic impact on the network .
the debtrank of @xmath58 containing only the single node @xmath17 is @xmath61 the debtrank , as defined in [ debtrank ] , excludes the loss generated directly by the default of the node itself and measures only the impact on the rest of the system through default contagion .
for some purposes , however , it is useful to include the direct loss of a default of @xmath17 as well . the total loss caused by the set of nodes @xmath58 in distress at time @xmath59 ,
including the initial distress is @xmath62 the precise meaning of the debtrank allows us to define the _ expected systemic loss _ for the entire economy @xcite . assuming that we have @xmath3 banks in the system , the expected systemic loss can be approximated by @xmath63 with @xmath19 the probability of default of node @xmath17 , and @xmath64 the combined economic value of all nodes at time @xmath26 . for details and the derivation ,
see @xcite . to calculate the marginal contributions to the expected systemic loss ,
we start by defining the _ net liability network _ @xmath65 $ ] .
after we add a specific liability @xmath66 , we denote the liability network by @xmath67 where @xmath68 is the kronecker symbol . the marginal contribution of the specific liability @xmath66 on the expected systemic loss is @xmath69 where @xmath70 is the debtrank of the liability network and @xmath71 the total economic value with the added liability @xmath66 .
clearly , a positive @xmath72 means that @xmath66 increases the total sr .
finally , the marginal contribution of a single loan ( or a transaction leading to that loan ) can be calculated .
we denote a loan of bank @xmath17 to bank @xmath1 by @xmath73 . the liability network changes to @xmath74 since @xmath17 and @xmath1 can have a number of loans at a given time @xmath26 , the index @xmath75 numbers a specific loan between @xmath17 and @xmath1 .
the marginal contribution of a single loan ( transaction ) @xmath76 , is obtained by substituting @xmath77 by @xmath78 in [ marginal_effect ] . in this way
every existing loan in the financial system , as well as every hypothetical one , can be evaluated with respect to its marginal contribution to overall sr .
the central idea of the srt is to tax every transaction between any two counterparties that increases sr in the system @xcite .
the size of the tax is proportional to the increase of the expected systemic loss that this transaction adds to the system as seen at time @xmath26 .
the srt for a transaction @xmath79 between two banks @xmath17 and @xmath1 is given by @xmath80 \quad .
\label{srteq_simple } \end{gathered}\ ] ] @xmath81 is a proportionality constant that specifies how much of the generated expected systemic loss is taxed .
@xmath82 means that 100% of the expected systemic loss will be charged .
@xmath83 means that only a fraction of the true sr increase is added on to the tax due from the institution responsible . for details , see @xcite . using debtrank
we can also define an _
average vulnerability _ of a node in the liability network @xmath36 @xcite . recall that @xmath84 is the state variable , which describes the health of a node in terms of equity capital after @xmath85 time steps ( @xmath86 , i.e. all nodes are either undistressed or inactive ) .
when calculating @xmath84 for @xmath87 containing only a single node @xmath17 , we can define @xmath88 as the health of bank @xmath1 in case of default of bank @xmath17 @xmath89 note that is necessary to simulate the default of every single node of the liability network @xmath36 separately to obtain @xmath90 .
with @xmath90 we can define the average vulnerability of a node in the liability network as the average impact of other nodes on node @xmath17 @xmath91 where @xmath3 is the number of nodes in the liability network @xmath36 .
we use the following three observables : ( 1 ) the size of the cascade , @xmath92 as the number of defaulting banks triggered by an initial bank default ( @xmath93 ) , ( 2 ) the total losses to banks following a default or cascade of defaults , @xmath94 , where @xmath7 is the set of defaulting banks , and ( 3 ) the average transaction volume in the interbank market in simulation runs longer than @xmath95 time steps , @xmath96 where @xmath97 represents new interbank loans at time step @xmath26 .
monica billio , mila getmansky , andrew w lo , and loriana pelizzon .
econometric measures of connectedness and systemic risk in the finance and insurance sectors .
_ journal of financial economics _ , 1040 ( 3):0 535559 , 2012
. sheri markose , simone giansante , and ali rais shaghaghi . financial network of us cds market : topological fragility and systemic risk . _ journal of economic behavior & organization _ , 830 ( 3):0 627646 , 2012 .
sebastian poledna , stefan thurner , j. doyne farmer , and john geanakoplos .
leverage - induced systemic risk under basle ii and other credit risk policies .
_ journal of banking & finance _ , 42:0 199212 , 2014 .
sebastian poledna , jos luis molina - borboa , serafn martnez - jaramillo , marco van der leij , and stefan thurner .
the multi - layer network nature of systemic risk and its implications for the costs of financial crises .
_ journal of financial stability _ , 20:0 7081 , 2015 .
domenico delli gatti , mauro gallegati , bruce c greenwald , alberto russo , and joseph e stiglitz .
business fluctuations and bankruptcy avalanches in an evolving network economy .
_ journal of economic interaction and coordination _ , 40 ( 2):0 195212 , 2009 .
stefano battiston , domenico delli gatti , mauro gallegati , bruce greenwald , and joseph e stiglitz .
liaisons dangereuses : increasing connectivity , risk sharing , and systemic risk .
_ journal of economic dynamics and control _ , 360 ( 8):0 11211141 , 2012 . d. delli gatti , e. gaffeo , m. gallegati , g. giulioni , and a. palestrini .
_ emergent macroeconomics : an agent - based approach to business fluctuations_. new economic windows .
springer , 2008 .
isbn 9788847007253 .
rama cont , amal moussa , and edson santos . network structure and systemic risk in banking systems . in jean - pierre fouque and joseph a langsam ,
editors , _ handbook of systemic risk _ , pages 327368 .
cambridge university press , 2013 .
stefano battiston , michelangelo puliga , rahul kaushik , paolo tasca , and guido caldarelli .
debtrank : too central to fail ? financial networks , the fed and systemic risk .
_ scientific reports _ , 20 ( 541 ) , 2012 . | in addition to constraining bilateral exposures of financial institutions , there are essentially two options for future financial regulation of systemic risk ( sr ) : first , financial regulation could attempt to reduce the financial fragility of global or domestic systemically important financial institutions ( g - sibs or d - sibs ) , as for instance proposed in basel iii .
second , future financial regulation could attempt strengthening the financial system as a whole .
this can be achieved by re - shaping the topology of financial networks .
we use an agent - based model ( abm ) of a financial system and the real economy to study and compare the consequences of these two options . by conducting three `` computer experiments '' with the abm we find that re - shaping financial networks
is more effective and efficient than reducing leverage . capital surcharges for g - sibs can reduce sr , but must be larger than those specified in basel iii in order to have a measurable impact .
this can cause a loss of efficiency .
basel iii capital surcharges for g - sibs can have pro - cyclical side effects . |
fu orionis systems ( hereafter fuors ) , named after their prototype object , are low - mass ( sun - like ) protostars that undergo a rapid accretion episode in the innermost region of the circumstellar disc ( see @xcite for a review ) .
the gravitational energy released in such an event leads to an emission outburst ( with a rise time of @xmath2yr and a duration of @xmath3yr ) , during which the disc is more luminous than the central protostar by a factor of @xmath4 .
the inferred mass accretion rate during the outburst is @xmath5 , much higher than typical accretion rates during the quiescent phase .
statistical arguments , first advanced by @xcite , indicate that such outbursts occur , on average , ten or more times during the protostellar lifetime .
there is evidence that the outbursts are more frequent during the early ( the so - called class 0 and class i ) evolutionary phases and peter out as the mass accretion rate declines and the protostar enters the visible ( class ii , or classical t tauri ) phase .
the picture that has emerged from the observations and their interpretation is that most of the mass that ends up in the protostar is transferred from the disc during such outbursts ( e.g. @xcite ; see , however , @xcite for a different viewpoint ) .
if our understanding is correct , the fu orionis phenomenon represents a key element of the star formation process .
fuor outbursts are accompanied by strong winds of maximum line - of - sight speeds @xmath6 ( e.g. * ? ? ?
* ) , whose inferred mass outflow rates @xmath7 can reach @xmath8 ( e.g. @xcite ; @xcite , hereafter chk93 ) .
if most of the mass accreted through the disc is indeed processed through outbursts of this type then most of the mass and momentum ejected over the protostellar lifetime and hence most of the impact that protostellar outflows may have on their environment ( e.g. in contributing to the dynamical support of the parent cloud against gravitational collapse and to the regulation of the mass inflow to the centre ) will be associated with these eruptions .
it is also likely that the repeated , powerful ejections have a strong influence on the properties and appearance of the large - scale jets that emanate from these protostars ( e.g. * ? ? ? * ) .
it is therefore important to understand the nature and origin of these outflows .
neither thermal nor radiative acceleration is likely to be important in fuor winds , which leaves magnetic driving as the most promising mechanism .
this conclusion is supported by a zeeman - signature least - square deconvolution measurement in the prototype object fu ori , which was interpreted as indicating the presence of a @xmath9 poloidal magnetic field on scales of @xmath10au @xcite .
one possible scenario is that the outflows represent a centrifugally driven wind that is launched along magnetic field lines that thread the disc and are sufficiently inclined ( at an angle @xmath11 for a keplerian rotation law ) to the disc surface ( e.g. * ? ? ?
these field lines could correspond to the interstellar magnetic field that threads the natal molecular cloud core and is dragged in by the accretion flow , although an origin in a disc dynamo is also conceivable . in this picture , the outflow need not be launched from the immediate vicinity of the central star .
alternatively , the wind , while still comprising material removed from the accretion disc , could be driven along stellar magnetic field lines . in view of the fact that the massive accretion flow during an fuor outburst is expected to compress the stellar magnetosphere to an equatorial radius @xmath12 not much larger than the stellar radius @xmath13
, the wind in this scenario necessarily originates close to the stellar surface .
one version of the latter scenario , proposed by @xcite , corresponds to the x - celerator model presented in @xcite . in this picture , the accretion flow spins up the outer layers of the star to breakup , resulting in a magnetocentrifugal wind being driven along a narrow bundle of opened - up field lines that emerge from an ` x - point ' at the stellar equator . in the x - wind model described in @xcite , the ` x - point ' is associated more generally with the corotation radius @xmath14 ( the radius where the disc angular speed equals the stellar angular speed ) , which in quiescent protostars is typically a few stellar radii away from the stellar surface .
chk93 and @xcite presented arguments in favour of the ` disc field ' interpretation of fuor outflows and against a ` stellar field ' scenario .
they demonstrated that increasingly stronger photospheric lines observed in fu ori become progressively more blueshifted even as their two absorption components ( attributed to the disc rotation ) move closer together in wavelength , and pointed out that this is precisely the behaviour expected in a disc - driven wind .
in addition , based on the evidence that most of the optical continuum in fuors is emitted by an extended disc and on the fact that typical stellar winds accelerate on scales comparable to the stellar radius , they contended that a stellar field - driven wind launched from the stellar surface could not reproduce the observations .
in particular , they argued that the strong , but only moderately blueshifted , intermediate - strength lines detected in fu ori could not originate in such a wind because , by the time such an outflow covered a significant fraction of the continuum emission region in the disc ( necessary for producing strong absorption ) , it would have already attained a high velocity ( and would therefore exhibit strongly blueshifted lines ) .
another potential problem that they cited involves the large rotational broadening that could be expected from a source rotating at breakup ( as envisioned in the x - celerator picture ) .
recent axisymmetric and 3d numerical simulations of disc accretion on to stellar magnetospheres ( @xcite , hereafter r09 ) have revealed features that resemble the x - wind configuration proposed by @xcite but that are nevertheless different on several counts .
specifically , it was found that such systems drive conical disc winds along stellar field lines that are bunched up by the accretion flow .
however , even though these winds also originate in the inner disc , their launching region is not confined to the immediate vicinity of the corotation radius , as hypothesized in the x - wind scenario .
furthermore , the conical winds are driven by the pressure gradient of the azimuthal magnetic field component ( wound up by the differential rotation between the disc and the star ) rather than centrifugally , and they have a smaller opening angle and a narrower lateral extent than x - winds .
interestingly , even though r09 only presented results for model parameters appropriate to protostars with comparatively low ( @xmath15 ) accretion rates , the outflows produced in their simulations exhibited several properties that could potentially mitigate the aforementioned arguments against stellar field - driven outflow models for fuors . in particular , it was found that the acceleration of a conical wind is more extended than in a typical ( hydrodynamic ) stellar outflow and that its rotation speed generally decreases along the flow , in contrast with the initial behaviour of a centrifugally driven wind .
these findings provide a strong motivation for reevaluating the viability of the ` stellar field ' class of wind models for fuors .
more recent investigations , employing larger simulation regions and higher accretion rates , have begun to extend the results of r09 .
one notable finding of this new work , analysed in @xcite , is that the collimation of conical winds increases with distance from the origin and that they can eventually become fully collimated . in this paper
we focus on simulations that we performed for parameters that are relevant to fuors .
our goal is to verify that conical winds are still produced under these circumstances and to examine whether they could potentially account for the inferred properties of fuor outflows . in section [ sec : model ]
we provide analytic estimates that are used to guide our simulations and we summarize our numerical scheme . in section [ sec : results ] we present representative results and derive the physical properties of the simulated flows .
we discuss the implications of this study for fuors in section [ sec : analysis ] and give our conclusions in section [ sec : conclude ] , where we also outline steps toward further progress .
the fuor phenomenon has been convincingly argued to represent an enhanced accretion episode in a protostellar accretion disc , most likely associated with an instability that arises from a mismatch between the mass accretion rates in the inner and outer disc regions ( e.g. @xcite , @xcite and references therein ) .
accordingly , we set up a numerical model that simulates a non - steady disc accretion ` burst ' on to a magnetized star .
we first discuss some basic scaling relations that allow us to choose the appropriate model parameters , and then briefly describe our numerical model .
our model is based on the assumption that the stellar magnetic field can effectively diffuse into the inner region of the disc , allowing the bulk of the inflowing disc material to be channelled on to the stellar surface along closed magnetic field lines ( e.g. * ? ? ?
* ; * ? ? ?
* ) and the remainder to be expelled in an outflow along opened - up field lines ( e.g. * ? ? ?
* ; * ? ? ?
the disc truncation ( or ` magnetospheric ' ) radius @xmath1 corresponds to the location in the disc where the torque exerted on the disc plasma by the stellar magnetic field becomes large enough to brake the disc keplerian rotation and enforce corotation with the star . for an aligned dipolar field ,
it is given by @xmath16 where @xmath17 is the stellar mass and @xmath18 is its magnetic dipole moment , @xmath19 is the gravitational constant and the mass accretion rate is measured at @xmath1 @xcite . under stationary conditions ,
the numerical factor @xmath20 is estimated to be @xmath21 @xcite .
when @xmath1 is close to @xmath13 , as in the fuor case , higher order magnetic moments @xmath22 , which produce magnetic field amplitudes @xmath23 , can also be expected to play a role . in this case ,
equation ( [ eq : r_m ] ) generalizes to @xmath24 @xcite , where , in particular , @xmath25 , 2 and 3 correspond , respectively , to the dipole , quadrupole and octupole field components . for a given dipole component , the incorporation of additional multipole components will tend to increase the value of the disc truncation radius over the estimate ( [ eq : r_m ] ) .
however , for the sake of simplicity , we restrict the discussion in the rest of this paper to a purely dipolar field .
in choosing our model parameters , we adopt as fiducial values the physical parameters inferred from observations of fu ori . in particular , based on the results given in @xcite , we take @xmath26 and @xmath27 .
these authors also estimate , from spectral modeling , that the inner radius @xmath28 of the fu ori disc is @xmath29 .
we identify this radius with @xmath1 , which allows us , using equation ( [ eq : r_m ] ) , to infer the value of @xmath30 : @xmath31 where the normalization of the equatorial surface magnetic field @xmath32 is consistent with typical values inferred in quiescent class - i @xcite and class - ii ( e.g. * ? ? ?
* ) protostars as well as with the results reported by @xcite for the poloidal field near fu ori . is larger than that of the corresponding unperturbed dipole field on account of the compressional amplification of the stellar field by the accretion flow . ]
the contribution of higher order multipole field components , which could become important near the stellar surface , would have the effect of increasing this ratio . ) and setting , for definiteness , @xmath33 , we deduce , using equation ( [ eq : r_m_n ] ) , that @xmath34 and @xmath35 . note in this connection that recent spectropolarimetric observations of the t tauri stars v2129 oph and bp tau ( @xcite , @xcite ; see also @xcite and @xcite ) inferred an octupolar surface field component that is stronger than the dipolar component . ]
previous accretion disc models of fuors have generally ignored the role of the magnetic field in truncating the disc and therefore identified @xmath28 , the inner radius of the disc , with the stellar radius ( e.g. * ? ? ?
the value @xmath36 inferred in this way is measurably higher than typical values for low - mass protostars ( @xmath37 ) , and several explanations have been advanced to account for the difference .
( a brief summary of this issue is given by @xcite , who favour an interpretation that attributes the larger radius to stellar expansion brought about by the deposition of heat produced by the accreting gas . ) since @xmath30 is generally @xmath38 in the magnetic accretion model ( for example , it is 1.4 in the representative simulation presented in this paper ) , the inferred value of @xmath13 is lower in this case , which reduces the implied difference from the radii of quiescent protostars .
we employ the same numerical model as the one described in section 2 of r09 , and the reader is referred to that paper for further details .
as explained in section 2.3.1 of r09 , the high - density gas that comprises the disc material enters the simulation region through the disc boundary only after the computation commences , and it subsequently flows inward on account of its viscosity .
this numerical setup is thus naturally suited for modelling the evolution of an accretion ` burst ' , which is the focus of the present work .
although we performed simulations for a variety of model parameters , we present only one representative case in this paper .
we use the same parameters as in the reference simulation shown in r09 , except that we reduce the reference radius @xmath39 from @xmath40 to @xmath41 and change the outer radius of the computational domain from @xmath42 to @xmath43.au , which corresponds to @xmath44 for the inferred value of @xmath13 in our reference simulation . ]
the latter change enables us to increase the mass accretion rate on to the central star in our simulation to the level inferred in fuors .
the larger size of the computational domain also allows us to get a better handle on the collimation properties of conical winds .
, is presented in @xcite .
] the mass accretion rate on to the central object is determined from the simulations using the expression @xmath45 where @xmath46 is the reference mass accretion rate and where @xmath47 and @xmath48 are , respectively , the dimensionless magnetic moment and mass accretion rate parameters .
we use @xmath49 as in the reference simulation of r09 and obtain the value of @xmath50 from the numerical calculation . as described in section [ sec : results ] , the final ( quasi - steady ) mass accretion rate on to the central object in our representative simulation corresponds to @xmath51 .
we also find that @xmath52 at that stage , which implies @xmath53 ( using our fiducial value for @xmath28 ) .
we can then use equation ( [ eq : mdot_sim ] ) to infer the value of @xmath54 for our representative model : @xmath55 the reference and fiducial parameters for our model are summarized in table [ tab : refval ] . ) , we infer @xmath56 . although @xmath20 is expected to be @xmath57 under stationary conditions ( e.g. * ? ? ?
* ) , we consider the derived value to be physically consistent , especially in view of the time - dependent nature of the simulation and the expected presence of higher order multipole field components .
one could in principle obtain a lower value of @xmath20 by increasing the adopted value of the parameter @xmath58 . ] [ cols="<,<,<,<",options="header " , ] the time evolution of the simulated system depends on the magnitudes of the viscosity and the magnetic diffusivity , which are parametrized by @xmath59 and @xmath60 , respectively .
r09 ( see their appendix d ) found that conical winds are established only when @xmath61 .
this is consistent with the fact that the dragging of the magnetic field by the accretion flow , which in fuors causes the field compression near the inner boundary of the disc , requires the magnetic prandtl number @xmath62 to be @xmath38 ( e.g. * ? ? ?
our simulations employ the values adopted in the reference case of r09 , namely @xmath63 and @xmath64 .
recent work on non - steady protostellar accretion - disc models ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) has indicated that @xmath59 must be large enough ( @xmath65 ) for outbursts that resemble those of fuors to be produced .
our adopted value of the viscosity parameter is consistent with this requirement .
it is noteworthy in this connection that r09 found that the formation of a robust conical outflow also requires @xmath59 ( _ and _ @xmath60 ) to be comparatively large ( @xmath66 ) .
the large - scale poloidal structure of the simulated flow is shown in fig . [
fig : fig1 ] at the time when the wind has become fully developed .
the most striking feature of the figure is its qualitative similarity to fig . 3 in r09 , which corresponds to a quiescent protostar that accretes at a much lower ( by a factor @xmath67 ) rate .
this conclusion is reinforced by an inspection of fig .
[ fig : fig2 ] , which shows a close - up view of the region near the star . in both cases a high - density , conical disc wind is launched from the vicinity of the disc truncation radius @xmath28 , and a lower - density , higher - velocity jet component is established in the interior of the cone . the main difference between the two simulations is in the value of @xmath68 : it is @xmath69 in our simulation , as compared with @xmath70 in the r09 reference calculation .
this difference is consistent with the expectation from equation ( [ eq : ratio ] ) , which indicates that this ratio scales only as a weak power of the mass accretion rate ( @xmath71 ) .
thus , even though the accretion flow is much more powerful in this case , the steep radial scaling of the magnetic pressure exerted by the dipolar field component ( @xmath72 ) insures that the disc is still truncated at a finite radius and does not actually ` crush ' the stellar magnetosphere ( as envisioned , for example , in the x - celerator scenario for fuors ; see @xcite ) .
although the disc in the current simulation is truncated very close to the stellar surface , the magnetic field structure in the vicinity of its inner radius is qualitatively very similar to the case where @xmath73 is @xmath38 .
in particular , the magnetic field lines that guide the conical wind and the axial jet are open .
the opening - up of parts of the initially dipolar stellar field is a consequence of the differential rotation between the star , where the magnetic field is anchored , and the disc , into which the field lines diffuse , and is a generic property of magnetically linked star disc systems ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . as in the lower-@xmath74 case considered in r09
, the conical wind is driven primarily by the pressure gradient of the azimuthal magnetic field component generated by the differential rotation rather than centrifugally .
note in this connection that the profile of the azimuthal velocity component ( see panel c in fig .
6 of r09 as well as the right panel of fig .
[ fig : fig4 ] below ) initially _ decreases _ along the poloidal field lines , which contrasts with the behaviour of centrifugally driven winds , in which the azimuthal speed @xmath75 at first increases along a field line . the acceleration is quite efficient , and the conical wind reaches outflow speeds @xmath76 ( corresponding to @xmath77 of the keplerian speed at @xmath28 ) at the outer edge of the simulation region . the highest poloidal speed observed in this simulation
is attained further up and is associated with the lower - density axial flow . as seen in fig .
[ fig : fig1 ] , its value is @xmath78 , which is consistent with the maximum line - of - sight speed measured in fu ori @xcite .
however , this value may not be accurate since the low - density region in the vicinity of the axis is susceptible to numerical artifacts . ) , which results in a comparatively weak axial outflow . ] near the base of the flow the wind velocity is dominated by the azimuthal component , which arises from the rotational motion of the disc in the wind - launching region and has a maximum value of @xmath79 , attained at @xmath80 .
( at smaller radii the wind azimuthal velocity decreases on account of the interaction with the magnetosphere , which rotates with the comparatively low angular velocity of the star . )
we note that , even though the wind is launched very close to the stellar surface and has a high initial rotation velocity , it does _ not _ originate in the stellar surface and does _ not _ require the star to rotate at break - up speed which distinguishes it from the outflow envisioned in the x - celerator scenario @xcite .
as discussed in r09 , the magnetic force also has a component directed toward the symmetry axis , which acts to collimate the wind . by using the poloidal matter flux distribution
, r09 determined that the conical outflow in their reference simulation attained an opening half - angle of @xmath81 . from the corresponding distribution presented in fig .
[ fig : fig1 ] , we find that the collimation is even more efficient in the case that we simulate , with the outflow half - angle decreasing to @xmath82 within a radial distance ( projected on the equatorial plane ) of @xmath83 from the stellar surface .
in general , a magnetically driven outflow is collimated by a combination of two effects ( e.g. * ? ? ?
* ) : the magnetic tension force that acts to balance the magnetic pressure - gradient force in the force - free sub - alfvnic regime , and the hoop stress exerted by the azimuthal magnetic field component in the super - alfvnic flow region .
the difference in the collimation properties of the conical wind in our simulation and in the reference simulation of r09 can be attributed to the fact that a higher mass accretion rate in the disc results in a stronger compression of the stellar magnetic field and hence in a larger collimating magnetic tension force in the sub - alfvnic region of the wind .
the hoop - stress effect is also stronger in the higher-@xmath74 case on account of the compressional amplification of the field and because the differential rotation that twists the field lines gets stronger ( for a given value of @xmath14 ) as @xmath28 is decreased .
a detailed analysis of the collimation properties of a conical wind from a high-@xmath74 disc is given in @xcite .
[ fig : fig3 ] shows the evolution of the matter fluxes that are deposited by the accretion flow on to the stellar surface and in the outflow ( with the mass outflow rate evaluated over a spherical surface far enough from the centre ) .
it is seen that the mean mass accretion rate increases steadily until it attains a fully developed state ( with @xmath74 corresponding to the value inferred in fu ori ) at a time @xmath84d from the start of the simulation . this time is longer than the @xmath85yr observed rise time of the fu ori outburst ( e.g. * ? ? ?
* ) , but the discrepancy is probably in large part just a consequence of the particular choice of initial conditions for our simulation ( see section [ subsec : numerical ] ) . in the fully developed state ,
the average accretion and outflow rates are related by @xmath86 .
this result is consistent with the observational findings in fu ori ( e.g. * ? ? ?
the rapid accretion during the fuor outburst can be expected to spin up the star , and it is therefore necessary to check whether our assumption of slow stellar rotation is self - consistent .
we calculated the torque on the star at the end of our simulation from the expression @xmath87 , where the reference torque @xmath88 is listed in table [ tab : refval ] and where the dimensionless field and matter contributions @xmath89 and @xmath90 are , in our case , @xmath91 and @xmath92 , respectively . and @xmath93 at @xmath94 at the end of the simulation .
this is consistent with the fact that , as in the quiescent case ( see fig . 14 in r09 )
, most of the angular momentum transport occurs through the viscous stress in the disc . ] by dividing the total torque calculated in this way , @xmath95 , into the stellar angular momentum @xmath96 , where we assume uniform rotation with angular velocity @xmath97 and scale the normalized radius of gyration @xmath98 by its value for a polytrope of index 1.5 , we infer a characteristic spin - up time @xmath99 , which is of the order of the typical fuor outburst time .
this implies that our assumption of a slow rotator is only marginally consistent .
we note , however , that a conical wind - like component may be present in the outflow even if this assumption is violated ( see section [ sec : conclude ] ) .
the simulation results presented in section [ sec : results ] indicate that the observed properties of fuor outflows could in principle be explained in terms of the conical wind and axial jet that are driven from the vicinity of the stellar surface in these systems along stellar magnetic field lines that are compressed , twisted , and opened up by the interaction between the initially dipolar field component and the strong accretion flow .
in particular , for typical values of the stellar mass , radius and surface magnetic field strength , and of the mass inflow rate at the inner edge of the circumstellar disc , our representative simulation demonstrates that this interaction can produce outflows whose properties ( mass outflow rate , strong rotational velocity component near the base and possibly also the maximum outflow speed ) are consistent with the observations . as was mentioned in section [ sec : intro ] , chk93 and @xcite argued against a stellar magnetic field - driven wind being able to account for the spectral properties of the outflow in fu ori .
they envisioned the outflow as originating in the stellar surface and accelerating rapidly along strongly divergent field lines even as its rotation speed ( which initially has the stellar breakup value ) continues to increase . in our picture , the absorption features modelled in the above - mentioned papers would arise in the conical - wind component , which exhibits a spatially more extended acceleration ( along fast collimating magnetic field lines ) and a lower initial rotation speed ( that at first actually decreases along the flow ) than the ` stellar field ' outflow assumed in those papers .
although the spatial and kinematic properties of our simulated conical wind are distinct from the semi - analytic disc outflow model presented in chk93 ( which combined a hydrostatic disc atmosphere with a simple representation of a centrifugal wind ) , it is probably qualitatively closer to that model ( which chk93 and hartmann & calvet 1995 argued was consistent with the spectral data for fu ori ) than to their hypothesized stellar wind model . to demonstrate that the conical wind model could account for the behaviour of the absorption lines measured in fu ori would require a determination of the thermal structure of the simulated flow and a calculation of the synthesized spectra of the relevant photospheric lines .
an analogous radiative - transfer calculation , addressing the rotationally induced line variability from an accreting t tauri star with a misaligned magnetic dipole , was carried out by @xcite . while a detailed computation of this type is outside the scope of the present paper
, we can obtain some indication of the potential promise of this model by calculating the density and velocity profiles as functions of distance from the mid - plane at the location of the optical continuum emission region and comparing the results with those obtained in the disc wind model of chk93 .
in view of the clear differences between our numerical model and chk93 s semi - analytic model ( which include the fact that the latter model , in contrast with our simulations , incorporates energy loss by radiative diffusion in the disc atmosphere ) , we can not expect to find a full quantitative correspondence between the two calculations . , @xmath100 and @xmath101 , and presented results for @xmath102 .
furthermore , in their spectral fits they assumed a disc inclination angle @xmath103 , whereas more recent fits for this source ( e.g. zhu et al .
2007 ) have used @xmath104 .
] however , we can look for common trends in the respective profiles .
we take the optical emission radius @xmath105 to correspond to an effective disc temperature of @xmath106k , and we use the disc model of @xcite , in which @xmath107 $ ] and the maximum effective temperature ( attained at @xmath108 ) is @xmath109k , to deduce @xmath110 . as noted by chk93 , the disc model fits of @xcite similarly imply that roughly 60% of the optical spectrum in fu ori arises from disc annuli between @xmath111 and @xmath112 .
chk93 suggested that this region could be adequately represented by their calculated ` disc atmosphere plus wind ' structure at @xmath113 , which they presented as a function of the height @xmath114 above the mid - plane in their table 2 . in view of the different model setups and
adopted source parameters , there are several possible choices for the value of the wind - launching radius where we could compare our model results with the ones given in that table . for definiteness
, we opt to also use @xmath115 .
note , however , that this value corresponds to different radial distances in the two models : @xmath116 for our choice of parameters and @xmath117 for those employed by chk93 .
the comparison between the predictions of these two models is presented in fig .
[ fig : fig4 ] . in the chk93 model
, the poloidal velocity only has a vertical ( @xmath114 ) component and the azimuthal velocity is approximated as being constant with height and set equal to the keplerian speed at the base of the flow ( @xmath118 for their adopted parameters ) .
the curves showing @xmath119 and the mass density @xmath120 in that model are labelled by chk and plotted as a function of @xmath114 at @xmath115 using the data in table 2 of chk93 . in view of the conical shape of our simulated wind , we plot the poloidal and azimuthal velocity components as well as the density in the numerical model along a slightly inclined path ( represented by the heavy dash - dotted straight line in fig . [ fig : fig2 ] ) . and [ fig : fig2 ] . ]
it is seen that , as expected , the values of corresponding quantities at a given distance from the mid - plane can be significantly different for the two cases . reflects , in part , the fact that our simulated disc is hotter and therefore geometrically thicker than the chk93 model disc .
the latter model was chosen to be consistent with the minimum - temperature requirement obtained by @xcite for a viscous disc to generate fuor - type outbursts .
the hotter disc produced in our simulation is clearly also consistent with this condition .
] however , we also find that the basic velocity and density structure of the two models is very similar . in both cases ,
the flow is rotation - dominated as it emerges from the disc but eventually @xmath121 comes to exceed @xmath75 ( which , in turn , does not systematically increase along the flow ) .
chk93 estimated that the disc photosphere at @xmath122 occurs roughly where the density drops to @xmath123 and @xmath119 increases to @xmath124 .
( for comparison , the sound speed in the optical emission region is @xmath125 . ) in their model , @xmath119 becomes @xmath126 when the density drops to @xmath127 . in our simulation
, @xmath128 increases above @xmath124 also roughly when the density drops to @xmath123 , and @xmath121 comes to exceed @xmath75 when @xmath120 decreases by another two orders of magnitude .
this correspondence indicates that the observed dependence of photospheric absorption line profiles in fu ori on the line strength , which was successfully reproduced by the chk3 model ( see also * ? ? ?
* ) , is consistent with an origin in a stellar field - driven conical wind . even if more extensive and detailed calculations indicate that the observed behaviour of the photospheric absorption lines in fu ori can not be reproduced by a conical wind model (
because , for example , a wind of this type that is launched very close to the stellar surface collimates too rapidly for its acceleration region to intercept a line of sight to the optical continuum emission region for the inferred disc inclination angle , or if the maximum predicted outflow speed is too low ) , the results of our simulations suggest that a stellar field - driven outflow might still be an important ingredient of a comprehensive model of fuors .
in particular , such an outflow could still potentially account for much of the mass and momentum injected into the ambient medium in the course of an outburst and perhaps also for the highest measured velocities in the h@xmath129 , h@xmath130 and na i lines ( e.g. * ? ? ?
* ; * ? ? ?
* ) even if another outflow component ( in particular , a disc wind driven along non - stellar magnetic field lines , which were not included in our simulations ) gives rise to the observed photospheric lines .
this is because a strong conical wind and a fast , low - density jet appear to be generic features of the disc / stellar - field interaction under a wide range of conditions . on the other hand , if it can be demonstrated that these predicted outflow components are , in fact , absent during fuor outbursts , this would indicate that at least one of the underlying key assumptions of the model [ e.g. that the magnetic diffusivity in the inner disc does not exceed the viscosity ( @xmath131 ) but is nevertheless sufficiently large ( @xmath132 ) , or that the star possesses a sufficiently strong dipolar field component ( @xmath133 ) ] is not valid , which would also enhance our physical understanding of these systems .
we have presented numerical simulation results that support an interpretation of the powerful winds that accompany fuor outbursts in terms of stellar magnetic field - driven disc outflows . in this picture , the massive accretion flow that gives rise to an observed burst strongly compresses the stellar magnetic field lines , and the resulting magnetic stress truncates the accretion disc very close to the stellar surface .
some of the field lines diffuse into the disc and become twisted by the differential rotation between the disc and the star .
this twisting , in turn , opens up the field lines , and the pressure gradient associated with the azimuthal magnetic field component along the opened field drives a moderate - velocity , dense conical wind that emanates from the vicinity of the truncation radius as well as a high - velocity , tenuous axial jet ( whose properties , however , are less well determined in the simulation that we described ) .
the magnetic field also acts to collimate these outflow components .
the conical wind and axial jet appear to be generic features of the interaction between an accretion disc and a predominantly dipolar stellar field in cases where the effective viscosity @xmath59 and magnetic diffusivity @xmath60 satisfy @xmath134 and are both comparatively high ( @xmath66 ) .
these features were originally identified in simulations of protostars with low and moderate accretion rates ( r09 ) . the representative simulation presented in this paper verifies that the same type of outflow
is produced also when the accretion rate is as high as @xmath135 , the value inferred in the archetypal object fu ori .
our simulation implies that the disc truncation radius in this source , which was observationally determined to lie at a radius @xmath136 , corresponds to a distance of @xmath137 from the stellar surface , and that the surface magnetic field is @xmath138 , which is consistent with independent indications .
the mass outflow rate in the simulated outburst ( dominated by the conical wind ) is a factor @xmath139 of the mass accretion rate on to the star , and the maximum outflow velocity within the computational domain ( attained in the axial jet ) is @xmath140 ; these agree well with the observationally inferred values for fu ori .
an interpretation of fuor winds in terms of an accretion - disc outflow driven along stellar magnetic field lines was previously proposed by @xcite on the basis of the x - celerator model of @xcite . in this picture , the outflow is launched centrifugally from the surface of a star whose outer layers rotate at breakup speeds .
this contrasts with the conical - wind scenario , in which the star rotates comparatively slowly and the outflow originates at a finite distance from the star and is driven by the @xmath141 magnetic pressure gradient from the start .
@xcite generalized the x - celerator model to the case where the star rotates below breakup , corresponding to the corotation radius @xmath14 exceeding @xmath13 .
however , in their generalized ( x - wind ) model , @xmath14 still coincides with the magnetospheric radius @xmath1 , and the nature of the outflow from that region ( the x - point ) is qualitatively similar to that of the x - celerator model . our assumption in this paper that @xmath142 is plausible in view of the fact that a rapidly rotating protostar could be efficiently braked through a magnetic interaction with the disc during the relatively long quiescent phase ( e.g. * ? ? ?
* ; * ? ? ?
and while such a star would be spun up during the rapid accretion event comprising an fuor outburst ( see section [ sec : results ] ) , this need not result in the surface layers reaching breakup speeds .
( note in this connection that , even in the absence of a large - scale magnetic field coupling the disc and the star , the protostellar surface layers are not expected to reach breakup speeds during an outburst of this type ; e.g. @xcite ) .
chk93 and @xcite argued that a stellar magnetic field - driven outflow model of the x - celerator type is inconsistent with the detection of moderately blueshifted , intermediate - strength absorption lines in fu ori , which , they suggested , could be explained in terms of a disc outflow originating at a distance @xmath143 of a few stellar radii and driven along magnetic field lines that are not associated with the star . specifically , they showed that the observed line profiles could be reproduced by a model in which gas launched from a keplerian accretion disc gradually accelerates until the poloidal velocity component comes to exceed the azimuthal velocity component . in this paper
we have demonstrated that a conical wind naturally exhibits this behaviour since the outflow also starts with a predominantly azimuthal velocity component and eventually accelerates to @xmath144 . in particular
, we showed that the density and poloidal velocity profiles along a ray through the conical shell that intercepts the disc at the distance of the optical emission region closely match the corresponding profiles calculated in the disc - outflow model of chk93 , notwithstanding the different setups ( and even the fiducial parameter values ) employed in the two ( respectively , numerical and semi - analytic ) models .
the fact that the azimuthal velocity of the conical wind remains much lower than the breakup speed of the star and that , in contrast with the initial behaviour of @xmath75 in a centrifugally driven wind , it does not increase ( but , rather , decreases ) along the outflow , circumvents another objection that chk93 levelled at the x - celerator scenario .
we note in this connection that an outflow component resembling a conical wind , as well as a strong axial jet component , have been found in simulations of the disc
magnetosphere interaction in the ` propeller ' regime ( @xmath145 ) . based on the results presented in r09
, we expect such a flow to be more strongly influenced by the centrifugal force and less well collimated for given values of @xmath18 and @xmath74 than the conical wind we considered above .
however , r09 also found in the propeller case that the magnetic force remains important in driving the wind and that the azimuthal speed of the wind does not increase along a field line ( see their fig .
it is therefore conceivable that an outflow in this regime could also account for the observations , although this remains to be verified by an explicit simulation .
while the results presented in this paper are highly suggestive , a more detailed calculation ( involving the thermal and spectral properties of the outflow ) is required to evaluate the contribution of a stellar field - driven disc wind to the absorption - line spectrum in an object like fu ori .
it will also be useful to carry out additional simulations in order to further check the dependence of the results on the adopted initial mass and magnetic flux distributions . in particular , our assumption that initially there is no disc ( as compared to coronal ) gas in the simulation region is not realistic , and our current numerical setup also does not account for the possibility that some of the stellar magnetic field may have diffused into the disc before the onset of the outburst ( see * ? ? ? * ) . given the comparatively high value of the disc inclination angle ( @xmath104 ; @xcite ) adopted in recent studies of fu ori , it is conceivable that the conical wind model would not be able to reproduce the absorption - line profiles measured in this object if the source of the optical continuum is indeed a region of size @xmath146 in the disc .
in that case a disc outflow driven along a non - stellar magnetic field , as proposed by chk93 , might provide the dominant contribution to the absorption - line spectrum of fu ori , although a ` conical wind plus axial jet ' outflow could potentially still contribute to some of the observed properties of this object .
note , however , that if a _ stellar _ wind is also present ( see section 5.2 of r09 ) , it would have a decollimating effect on the conical wind ( e.g. * ? ? ?
* ; * ? ? ?
* ) that could increase the range of disc radii ` covered ' by the conical outflow . to our knowledge
, photospheric line shifts such as those detected in fu ori have so far not been found in any other fuor .
although a lower-@xmath147 wind or some other factor ( such as a higher projected azimuthal velocity ) could have prevented a detection in other fuors ( see * ? ? ?
* ) , it would clearly be useful to be able to test competing models also in other bursting sources .
lower - amplitude , repetitive photometric outbursts have been detected in ex lupi and a few other t tauri stars ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and they have also been interpreted as enhanced mass accretion events . in a few of these exor sources
there is evidence for an accompanying outflow , which , as discussed in r09 , may well ( at least in some cases ) arise in a disc magnetosphere interaction .
however , these systems generally do not exhibit absorption - line spectra like fuors , and there is also no indication that their continuum emission is dominated by a disc ; in this regard their appearance is similar to that of quiescent systems . therefore , unless other spectroscopic diagnostics are identified in these sources , fuors will remain the best candidates for probing the acceleration regions of protostellar disc outflows .
this research was supported in part by nsf grants ast-0908184 ( ak ) and ast-1008636 ( mmr and rvel ) and by a nasa atp grant nnx10af63 g ( mmr and rvel ) . the authors thank g. v. ustyugova and a. v. koldoba for the development of the code used in the simulations discussed in this paper , and the reviewer for comments that helped to improve the presentation .
bastian u. , mundt r. , 1985 , a&a , 144 , 57 blandford r. d. , payne d. g. , 1982 , mnras , 199 , 883 calvet n. , hartmann l. , kenyon s. j. , 1993 , apj , 402 , 623 ( chk93 ) calvet n. , hartmann l. , strom s. e. , 2000 , in mannings v. g. , boss a. p. , russell s. , eds , protostars & planets iv .
arizona press , tucson , p. 377
clarke c. j. , lin d. n. c. , pringle j. e. , 1990 , mnras , 242 , 439 croswell k. , hartmann l. , avrett e. h. , 1987 , apj , 312 , 227 donati j. f. , et al . , 2007 , mnras , 380 , 1297 donati j. f. , et al .
, 2008 , mnras , 386 , 1234 donati j. f. , paletou f. , bouvier j. , ferreira j. , 2005 , nature , 438 , 466 fendt c. , 2009 , apj , 692 , 346 ghosh p. , lamb f. k. , 1979a , apj , 232 , 259 ghosh p. , lamb f. k. , 1979b , apj , 234 , 296 goodson a. p. , winglee r. m. , 1999 , apj , 524 , 159 hartmann l. , calvet n. , 1995 , aj , 109 , 1846 hartmann l , kenyon s. j. , 1996 , ara&a , 34 , 207 herbig g. h. , 1977 , apj , 217 , 693 herbig g. h. , 1989 , in reipurth , b. , ed , eso workshop on low - mas star formation and pre main - sequence objects .
eso , garching , p. 233
herbig g. h. , 2007 , aj , 133 , 2679 herbig g. h. , 2008 , apj , 135 , 637 herbig g. h. , petrov p. p. , duemmler r. , 2003 , apj , 595 , 384 johns - krull c. m. , 2007 , apj , 664 , 975 johns - krull c. m. , greene t. p. , doppmann g. w. , covey k. r. , 2009 , apj , 700 , 1440 kenyon s. j. , hartmann l. , hewett r. , 1988 , apj , 325 , 231 knigl a. , 1991 , apj , 370 , l39 kurosawa r. , romanova m. m. , harries t. j. , 2008 , mnras , 385 , 1931 lii p. , romanova m. m. , lovelace r. v. e. , 2011 , mnras , submitted ( arxiv:1104.4374 ) long m. , romanova m. m. , kulkarni a. k. , donati j. f. , 2011 , mnras , 413 , 1061 long m. , romanova m. m. , lamb f. k. , 2011 , new astronomy , submitted ( arxiv:0911.5455 ) long m. , romanova m. m. , lovelace r. v. e. , 2005 , apj , 634 , 1214 lovelace r. v. e. , romanova m. m. , bisnovatyi - kogan g. s. , 1995 , mnras , 275 , 244 lubow s. h. , papaloizou j. c. b. , pringle , j. e. , 1994 , mnras , 267 , 235 lynden - bell d. , boily c. , 1994 , mnras , 267 , 146 malbet f. , et al . , 2005 , a&a , 437 , 627 meliani z. , casse f. , sauty c. , 2006 , a&a , 460 , 1 popham r. , kenyon s. , hartmann l. , narayan r. , 1993 , apj , 473 , 422 popham r. , narayan r. , hartmann l. , kenyon s. , 1993 , apj , 415 , l127 reipurth b. , 1990 , in mirzoyan , l. v. , pettersen b. r. , tsvetkov m. k. , eds , proc .
137 , kluwer , dordrecht , p. 229 romanova m. m. , long m. , lamb f. k. , kulkarni a. k. , donati j .- f . , 2011 ,
mnras , 411 , 915 romanova m. m. , ustyugova g. v. , koldoba a. v. , lovelace , r. v. e. , 2009 , mnras , 399 , 1802 ( r09 ) shu , f. h. , lizano s. , ruden s. p. , najita j. , 1988 , apj , 328 , l19 shu , f. najita j. , ostriker e. , wilkin f. , ruden s. , lizano s. , 1994 , apj , 429 , 781 skinner s. l. , sokal k. r. , gdel m. , briggs k. r. , 2009 , apj , 696 , 766 ustyugova g. v. , koldoba a. v. , romanova m. m. , lovelace r. v. e. , 2006 , apj , 646 , 304 uzdensky d. , knigl a. , litwin c. , 2002 , apj , 565 , 1191 van ballegooijen a. a. , 1994 , ssrv , 68 , 299 zhu z. , hartmann l. , calvet n. , hernandez j. , 2007 , apj , 669 , 483 zhu z. , hartmann l. , calvet n. , hernandez j. , tannirkulam a .- j . , dalessio p. , 2008 ,
apj , 684 , 1281 zhu z. , hartmann l. , gammie , c. , 2009a , apj , 694 , 1045 zhu z. , hartmann l. , gammie c. , book l. c. , simon j. b. , engelhard e. , 2010 , apj , 713 , 1134 zhu z. , hartmann l. , gammie c. , mckinney , j. c. , 2009b , apj , 701 , 620 | fu orionis ( fuor ) outbursts are major optical brightening episodes in low - mass protostars that evidently correspond to rapid mass - accretion events in the innermost region of a protostellar disc .
the outbursts are accompanied by strong outflows , with the inferred mass outflow rates reaching @xmath0 of the mass inflow rates .
shu et al . proposed that the outflows represent accreted disc material that is driven centrifugally from the spun - up surface layers of the protostar by the stellar magnetic field .
this model was critiqued by calvet et al . , who argued that it can not reproduce the photospheric absorption - line shifts observed in the prototype object fu ori .
calvet et al . proposed that the wind is launched , instead , from the surface of the disc on scales of a few stellar radii by a non - stellar magnetic field . in this paper we present results from numerical simulations of disc accretion on to a slowly rotating star with an aligned magnetic dipole moment that gives rise to a kilogauss - strength surface field .
we demonstrate that , for parameters appropriate to fu ori , such a system can develop a strong , collimated disc outflow of the type previously identified by romanova et al . in simulations of protostars with low and moderate accretion rates . at the high accretion rate that characterizes the fuor outburst phase , the radius @xmath1 at which the disc is truncated by the stellar magnetic field moves much closer to the stellar surface , but the basic properties of the outflow , which is launched from the vicinity of @xmath1 along opened - up stellar magnetic field lines , remain the same .
these properties are distinct from those of the x - celerator ( or the closely related x - wind ) mechanism proposed by shu et al . in particular
, the outflow is driven from the start by the magnetic pressure - gradient force , not centrifugally , and it is more strongly collimated .
we show that the simulated outflow can in principle account for the main observed characteristics of fuor winds , including the photospheric line shifts measured in fu ori .
a detailed radiative - transfer calculation is , however , required to confirm the latter result .
[ firstpage ] accretion , accretion discs mhd stars : formation stars : magnetic field stars : winds , outflows . |
the presence of fe k features in the spectra of agn and the interpretation on their origin are nowadays quite well established .
fe k@xmath0 transitions give rise to fluorescence emission lines .
the neutral fe k line at 6.4 kev is usually the most prominent in agn spectra . an unresolved component of this transition is almost ubiquitous in xmm - newton agn spectra ( bianchi et al .
2004 , page et al .
2004 , jimnez - bailn et al .
2005 ) and it is generally interpreted as being emitted via x - ray reflection on a distant medium , like the agn torus ( ghisellini et al .
1994 ) . in some cases
it is also possible to observe fe k lines from ionised iron atoms emitted at 6.7 and 6.97 kev ( see left panel of fig.1 for an example of fe k lines complex in the sy 1 mrk 590 ) . the origin of these lines is ascribed to warm gas surrounding the accretion disc . when fe k emission is produced in the innermost region of the agn ,
relativistic effects intervene and they can strongly modify the line profile , giving rise to a broad and skewed fe k line ( fabian et al .
this was the general picture up to few years ago .
in fact the debate on the nature of the inner accretion flow in agn has been reinforced by the recent discoveries of narrow energy - shifted spectral features in the spectra of several seyfert 1 galaxies .
narrow energy - shifted _ emission lines _ have been detected in the hard x - ray spectra of some agn : ngc 3516 ( turner et al .
2002 , iwasawa et al .
2004 ) , mrk 766 ( turner et al . 2004 , 2006 ) , eso 198-g24 ( guainazzi 2003 ) , ngc 7314 ( yaqoob et al .
2003 ) , mrk 841 ( petrucci et al . 2002 , longinotti et al .
2004 ) , eso 113-g010 ( porquet et al .
2004 ) , ugc 3973 , ( gallo et al .
an example of a redshifted emission line detected in ngc 3516 is shown in the right panel of fig .
theoretical models have predicted the possibility that fe k emission lines from the disc can be observed with a narrow profile under some conditions .
in fact , the broad fe line profile arises by integrating the emission from a large area of the disc .
but , if the x - ray reflection arises as a result of magnetic flares in localized regions on the disc , energy - shifted narrow lines are instead expected to be observed in the spectrum .
nayakshin et al .
( 2001 ) proposed a model in which the accretion disc is illuminated by magnetic flares inducing active regions ( hot spots ) with limited size .
as these regions orbit around the black hole , the emitted line should appear in a range of energies between @xmath1 48 kev due to the doppler and gravitational effects .
if the emitting region co - rotates with the disc , the flux and the centroid energy of the line vary as a function of the orbital phase . as demonstrated by dovciak et al .
( 2004 ) , the line profile computed in time - resolved intervals does show such variations and in theory , it is possible to reconstruct the path of the hotspot on the disc .
another scenario in which narrow energy - shifted lines can be produced has been proposed by turner et al .
( 2002 , 2004 ) .
redshifted lines are interpreted as being emitted by material which is either being ejected at relativistic velocity on the far side or falling towards the black hole .
the different ranges of energies of the line peaks would then reflect respectively a deceleration or acceleration in the velocity of the gas due to the gravitational force .
remarkably , the `` aborted jets '' model recently proposed by ghisellini et al .
( 2004 ) to explain the x - ray emission in radio - quiet agn , predicts the presence of blobs of gas in fast motion close to the black hole .
x - ray _ absorption lines _ in the fe k band may also provide insights on the central region of agn .
the first detection of an absorption redshifted fe k line was found in _
data of ngc3516 ( nandra et al . 1999 ) . with the advent of _
xmm@xmath2newton _ and _ chandra _ , the number of absorption features in active galaxies spectra have considerably increased : q0056 - 363 ( matt et al . 2005 ) , e @xmath3 ( yaqoob & serlemitsos 2005 , but see also jimnez - bailn et al . 2006 ) , mrk 509 ( dadina et al . 2005 ) , pg@xmath4 ( reeves et al . 2005 ) , mrk 335 ( longinotti et al .
one of the best example of variable absorption structures from k@xmath0 and k@xmath5 transitions of ionised iron is the xmm - newton spectrum of ngc 1365 ( risaliti et al . 2005 ) displayed in the left panel fig .
2 , whereas on the right side of the same figure is plotted the redshifted absorption line detected in mrk 335 ( see caption for details ) .
several cases of x - ray blueshifted absorption lines have been interpreted as evidence for high velocity outflows : apm 08279 + 5255 and pg1115 + 080
( chartas et al . 2002 , 2003 ) ,
pg1211 + 143 ( pounds et al . 2003 ) , pds 456 ( reeves et al .
2003 ) , although few of them somewhat questionable ( mckernan et al .
the outflows interpretation is also supported by theoretical models where synthetic spectra produced through radiative transfer calculations are found in good agreement with the observations on a qualitative level ( sim 2005 ) . according to accretion models
, high velocity gas should be present in the surroundings of the central region of agn and the presence of such gas is in principle observable in the form of absorption features imprinted on the x - ray spectra ( crenshaw et al .
depending on the kinematic structure and dynamical state of the gas , the absorption lines would be observed with different widths and with some displacement from their rest frame energy .
this gas is likely to be highly ionised due to the nuclear continuum irradiation and therefore is bound to produce absorption mainly in the fe k shell band .
resonance transitions from fexxv and fexxvi produce absorption features at the energy of 6.7 and 6.97 kev . taking into account the source redshift
, these features should be observed in the @xmath158 kev range for most objects . to be observed with considerable energy - shift
, the lines must be forming in a plasma infalling on the accretion disc or or outflowing at high velocity .
although the exact nature of the energy shift of such lines is as yet unclear , the most likely scenario for producing the observed features would involve a combination of gas orbiting in highly relativistic motion and/or gravitational shifts of the photons ( ruszkowski et al .
emission and absorption energy@xmath2shifted lines are clearly two distinct phenomena , but they share some common properties : * they are detected with marginal significance ( 2@xmath23 @xmath6 ) * they have been found in individual objects , by individual authors and therefore with various methods of data reduction and analysis , and with heterogeneous calibrations * they are probably of transient nature , since most detections have been found in time@xmath2resolved spectra . * high confidence detections of such features would be of crucial importance in testing the black hole paradigm for agn and would provide a new additional tool to be used alongside the broad fe k@xmath0 line .
the project described in the following aims to perform a systematic search on narrow energy - shifted lines in the _ xmm - newton _ archive and to characterise the significance of the detections with monte carlo simulations .
the data have been collected from the _ xmm - newton _ archive .
the spectra are part of a larger sample of agn including all the observations flagged as `` agn '' in the _ xmm - newton _ archive and public up to march 2006 ( bianchi et al . in preparation ) .
we have selected only the sources with number of counts in the hard x - ray band higher than 1000 counts , obtaining a list of 124 spectra extracted from 85 type-1 agns ( multiple observations are available for many sources ) .
table 1 reports the optical classification of the objects .
the data have been processed with sas version 6.5.0 and analysed with xspec 12.2.1ao .
the procedure described hereafter is adapted from longinotti et al .
( 2006b ) , where it was used to test the significance of the redshifted absorption line in mrk 335 . for the present work
, each spectrum in the sample has been fitted in the 2 - 10 kev range with a baseline model defined by a power law + 3 gaussian emission lines with energies at 6.4 , 6.7 , 6.97 kev and width fixed to 1 ev .
then , the presence of additional narrow lines is tested by adding _
n _ narrow gaussian lines to the baseline model , allowing positive and negative deviations for the line intensity and excluding the range 6.4 - 7 kev .
the energy of these lines is free to vary between 4 and 9 kev . in this way a best fit model made by a power law+3 fe k
lines + _ n _ gaussian lines is found .
the significance of these additional gaussian lines with respect to the best fit model is checked through the f - test .
only those lines that are found significant at more than 99% for the f - test are included in the best fit ( this conservative threshold has been arbitrarily chosen ) . in the next step
we test the significance of the _
n _ gaussian lines in the best fit .
a line detected in this way could be due to random deviations originated by poisson noise in the data .
thus , the question is `` what is the probability that a deviation in @xmath7 this large or larger is obtained by chance ? '' . to answer it
, we run monte carlo simulations on each spectrum where a shifted line has been detected at more than 99% for the f - test .
the algorithm for the simulations is as follows .
we produce 10@xmath8 realisations of the spectrum fitted with an input model made by the best fit model without the line that is being tested . in this way we obtain 10000 fake background - subtracted data sets with photon statistics and spectral shape corresponding to the original spectrum .
each of the simulated spectra is fitted with the input baseline model .
then , the gaussian line to be tested is added back to the baseline model , a new spectral fit is performed and the improvement in @xmath7 for adding that line is recorded for each of the simulated spectra .
the number of @xmath9 larger than the one found in the real spectrum will be used to estimate the significance of the line in the real spectrum . at the time of the writing ,
only few test - runs have been performed , setting the threshold of significance at 99 and 95% , in order to understand how the method works .
this allow us to identify some drawbacks of the method and improve it before reporting unclear or incomplete results .
some of these problems are highlighted in the following : the method does not take into account the complexity of many agn spectra , particularly in seyfert galaxies .
for example , if a broad line or spectral curvature is present , the automatic procedure would mimick the broad feature with many adjacent narrow lines .
the transient nature of the shifted features constitutes another issue to be taken into account and solved : in a systematic search on the integrated spectra , some of the transient feature would be lost , as it happens in our sample for ngc 3516 .
however , a preliminary check on the well known objects pg1211 + 143 and mrk 509 where detections of narrow lines have been claimed , confirms the presence of shifted lines and therefore we are confident that our procedure provides a good basis to search for shifted lines systematically . among the many future developments of this work
, we foresee to test the detection procedure varying the baseline models for the simulated spectra .
we would like to run it in samples of different objects in the _ xmm - newton _ archive ( e.g. blazars ) to be used as a control sample and to take into account the transient nature of the features performing the search in time - resolved spectra .
bianchi s. , matt g. , balestra i. , guainazzi m. , perola g. c. , 2004 , a&a , 422 , 65 chartas g. , brandt w. n. , gallagher s. c. , garmire g. p. , 2002 ,
apj , 579 , 169 chartas g. , brandt w. n. , gallagher s. c. , 2003 , apj , 595 , 85 crenshaw d. m. , kraemer s. b. , george i. m. , 2003 , ara&a , 41 , 117 dadina m. , cappi m. , malaguti g. , ponti g. , de rosa a. , 2005 , a&a , 442 , 461 doviak m. , bianchi s. , guainazzi m. , karas v. , matt g. , 2004 , mnras , 350 , 745 fabian a. c. , iwasawa k. , reynolds c. s. , young a. j. , 2000 , pasp , 112 , 1145 gallo l. c. , fabian a. c. , boller t. , pietsch w. , 2005 , mnras , 363 , 64 ghisellini , g. , haardt , f. , & matt , g. 1994 , mnras , 267 , 743 ghisellini , g. , haardt , f. , & matt , g. 2004 , a&a , 413 , 535 guainazzi m. , 2003 , a&a , 401 , 903 iwasawa k. , miniutti g. , fabian a. c. , 2004 , mnras , 355 , 1073 jimnez - bailn , e. , piconcelli , e. , guainazzi , m. , schartel , n. , rodrguez - pascual , p. m. , & santos - lle , m. 2005 , a&a , 435 , 449 jimnez - bailn , e. , santos - lle , m. , piconcelli , e. , matt , g. , guainazzi , m. , rodrguez - pascual , p. , astro - ph /0610233 , a&a in press longinotti a. l. , nandra k. , petrucci p. o. , oneill p. m. , 2004 , mnras , 355 , 929 mckernan b. , yaqoob t. , reynolds c. s. , 2005 , mnras , 658 nandra k. , george i. m. , mushotzky r. f. , turner t. j. , yaqoob t. , 1999 , apj , 523 , l17 nayakshin s. , kazanas d. , 2001 , apj , 553 , l141 page , k. l. , obrien , p. t. , reeves , j. n. , & turner , m. j. l. 2004 , mnras , 347 , 316 petrucci p. o. , et al . , 2002 , a&a , 388 , l5 porquet d. , reeves j. n. , uttley p. , turner t. j. , 2004 , a&a , 427 , 101 pounds k. a. , reeves j. n. , king a. r. , page k. l. , obrien p. t. , turner m. j. l. , 2003 , mnras , 345 , 705 reeves j. n. , pounds k. , uttley p. , kraemer s. , mushotzky r. , yaqoob t. , george i. m. , turner t. j. , 2005 , apj , 633 , l81 risaliti , g. , bianchi , s. , matt , g. , baldi , a. , elvis , m. , fabbiano , g. , & zezas , a. 2005 , apjl , 630 , l129 ruszkowski m. , fabian a. c. , 2000 , mnras , 315 , 223 sim s. a. , 2005 , mnras , 356 , 531 turner t. j. , et al . , 2002 , apj , 574 , l123 turner t. j. , kraemer s. b. , reeves j. n. , 2004 , apj , 603 , 62 turner , t. j. , miller , l. , george , i. m. , & reeves , j. n. 2006 , a&a , 445 , 59 yaqoob t. , padmanabhan u. , 2004 , apj , 604 , 63 yaqoob t. , serlemitsos p. , 2005
, apj , 623 , 112 | the detection of x - ray narrow spectral features in the 5 - 7 kev band is becoming increasingly more common in agn observations , thanks to the capabilities of current x - ray satellites .
such lines , both in emission and in absorption , are mostly interpreted as arising from iron atoms . when observed with some displacement from their rest frame position
, these lines carry the potential to study the motion of circumnuclear gas in agn , providing a diagnostic of the effects of the gravitational field of the central black hole .
these narrow features have been often found with marginal statistical significance .
we are carrying on a systematic search for narrow features using spectra of bright type1 agns available in the _ xmm - newton _ archive .
the aim of this work is to characterise the occurrence of the narrow features phenomenon on a large sample of objects and to estimate the significance of the features through monte carlo simulations .
the project and preliminary results are presented .
address = european space astronomy centre , apartado 50727 e-28080 , madrid , spain |
carrier propagation in mott or charge - transfer insulators is a challenging problem particularly motivated by strongly correlated superconducting cuprates @xcite .
while holes move incoherently in one - dimensional systems featuring charge and spin separation @xcite as well as in systems with antiferromagnetic ( af ) ising interactions @xcite , the hole motion becomes coherent and quasiparticles ( qps ) arise at low energy in the quantum af @xmath0-@xmath1 model @xcite .
these qps are indeed observed in angle - resolved photoemission ( arpes ) experiments in cuprates @xcite . in general , low - energy qps coexist with incoherent processes at high energy , as in the af phase on the square @xcite or honeycomb lattice @xcite .
this is however not always the case as shown by arpes experiments for the spin - orbit mott insulator na@xmath4iro@xmath5 @xcite , without clear evidence for qps at low energy @xcite . in a recent study hole - doped li@xmath4ir@xmath6ru@xmath7o@xmath5 with honeycomb structure
was found insulating at all doping levels @xcite .
electronic systems with honeycomb lattice include graphene @xcite , optical lattices @xcite , topological insulators @xcite , and frustrated magnets @xcite .
interest in the latter was triggered by theoretical predictions @xcite that na@xmath4iro@xmath5 may host kitaev model physics and quantum spin hall effect .
the ground state of this model is a kitaev spin liquid ( ksl ) characterized by finite spin correlations only for nearest neighbor ( nn ) spins @xcite .
the ksl belongs to the spin disordered phases which are in the center of interest in quantum magnetism @xcite . in mott insulators where the strong spin - orbit interaction generates a kramers doublet from partly filled @xmath8 orbitals @xcite , as in na@xmath4iro@xmath5 ,
effective @xmath9 pseudospins stand for local , spin - orbital entangled @xmath8 states which form orbital moments @xcite .
both kitaev and heisenberg interactions emerge from the spin and orbital coupling on the honeycomb lattice of iridium ions and form the kitaev - heisenberg ( kh ) model @xcite .
experimental observations revealed zigzag ( zz ) magnetic order in na@xmath4iro@xmath5 @xcite it may be explained within the kh model with next - nearest neighbor ( nnn ) @xmath10 and third nn ( 3nn ) @xmath11 af interactions @xcite .
the phase diagram of the frustrated heisenberg @xmath12-@xmath10-@xmath11 af model @xcite includes the nel af , zz , and also stripy ( st ) phase .
these phases survive when more general spin interactions with symmetric off - diagonal exchange are considered @xcite .
doping the ksl is particularly exciting as the ground state is spin disordered and it is unclear whether qps would form @xcite .
moreover in the ksl interactions may emerge that lead to unexpected forms of superconductivity .
indeed , slave - boson studies found here @xmath13-wave superconductivity at intermediate doping @xcite , whereas fermi liquid was claimed at light doping @xcite .
an important first step to explore hole doping is however the study of single hole motion and the existence of qps .
so far , spin - charge separation was shown for the kagome lattice , being a prototype of a spin liquid , while small qp peaks were found at some momenta in a frustrated checkerboard lattice @xcite .
the @xmath0-@xmath1-like kh model provides here a unique opportunity to investigate hole propagation in quite different magnetic phases emerging from frustrated interactions .
the purpose of this paper is : ( i ) to investigate the evolution of the spectral properties in the kh model under increasing frustration of magnetic interactions , ( ii ) to recognize the qp behavior in various magnetic phases of the frustrated kh model , and ( iii ) to establish whether the disordered ksl indeed realizes a paradigm of a fermi liquid at light doping as suggested in ref .
@xcite . in our study
we employ exact diagonalization ( ed ) of finite periodic clusters which has the important advantage that no approximations have to be made , which is particularly important as the analytical theory of hole - motion in the kh model is notoriously difficult and largely unexplored .
the possible problems with the ed approach are finite size effects and the difficulty to extrapolate to the thermodynamic limit ( tl ) . as important results ,
we report the spectral functions of the different ordered phases and their respective quasiparticle features . moreover we report a totally incoherent spectral weight distribution for a hole moving in the kitaev spin - liquid phase in the strong coupling regime , i.e. , relevant for the iridate systems .
the absence of qps in this case for finite systems clearly suggests that also in the tl there are no qps at strong coupling . for this reason
we conclude that then a dilute gas of holes ( that are individually not qps ) will not turn into a fermi liquid . the paper is organized as follows : in the sect .
ii we outline the kh model and discuss the calculation of various correlation functions that are used to determine the phase diagram of the kh - model .
section iii deals with the motion of holes in the ordered phases of the model and discusses the different propagators and spectral functions used . in sect .
iv we address the hole motion in the ksl and analyse the spectral weight distribution .
the latter is discussed in the intermediate and strong coupling limit .
moreover the dynamical spin structure factor is analysed for the ksl both for the kitaev and the kh model in order to explore the different scattering channels for holes .
results are summarized in sect . v.
we consider the following @xmath0-@xmath1-like kh model ( @xmath14 ) , on the honeycomb lattice [ fig .
[ fig : defs](a ) ] : @xmath15 it consists of the kinetic energy term @xmath16 of projected fermions with flavor @xmath17 @xcite which move in the restricted space without double occupancies as a result of large on - site coulomb repulsion @xmath18 .
the spins @xmath9 are defined in terms of fermionic creation ( annihilation ) operators @xmath19 ( @xmath20 ) with flavor @xmath17 at site @xmath21 : @xmath22 where @xmath23 are pauli matrices , @xmath24 .
we emphasize that already the kitaev terms @xmath25 with different ising spin interactions that depend on bond direction introduce strong spin frustration in eq . .
in the following we shall assume ferromagnetic ( fm ) kitaev ( @xmath26 ) and af heisenberg ( @xmath27 ) exchange , @xmath28 here @xmath1 is the energy unit and @xmath29 $ ] is a parameter that interpolates between the heisenberg and kitaev exchange couplings for nn spins @xmath9 .
the model eq .
( [ model ] ) includes nnn @xmath30 and 3nn @xmath31 af terms as well .
we use exact diagonalization ( ed ) within the lanczos algorithm for a periodic cluster of @xmath32 sites which accomodates all point group symmetries of the infinite lattice .
the momenta corresponding to allowed symmetry representations are presented in the first brillouin zone ( bz ) in fig .
[ fig : defs](b ) .
we introduce @xmath33 , @xmath34 , and @xmath35 , where @xmath36 and @xmath37 .
note that in absence of symmetry - breaking field , @xmath38 and @xmath39 are identical and only two distinct representations exist , @xmath40 .
sites and the elementary translations @xmath41 that connect atoms @xmath42 within one sublattice , i.e. , connect unit cells consisting of one atom @xmath42 and one @xmath43 each .
heisenberg couplings between nn ( @xmath12 ) , nnn ( @xmath10 ) and 3nn ( @xmath11 ) in @xmath44 ( [ model ] ) are indicated by solid and dashed lines ; kitaev couplings @xmath45 involve a single spin component for each bond direction , @xmath46 .
( b ) first bz with high symmetry @xmath47 and @xmath48 points ( in absence of symmetry breaking these points are equivalent to one another).,title="fig:",width=143 ] .0 cm sites and the elementary translations @xmath41 that connect atoms @xmath42 within one sublattice , i.e. , connect unit cells consisting of one atom @xmath42 and one @xmath43 each .
heisenberg couplings between nn ( @xmath12 ) , nnn ( @xmath10 ) and 3nn ( @xmath11 ) in @xmath44 ( [ model ] ) are indicated by solid and dashed lines ; kitaev couplings @xmath45 involve a single spin component for each bond direction , @xmath46 .
( b ) first bz with high symmetry @xmath47 and @xmath48 points ( in absence of symmetry breaking these points are equivalent to one another).,title="fig:",width=151 ] + in the present ed approach of finite systems there is no spontaneous symmetry breaking , and the spin components @xmath49 , @xmath50 , and @xmath51 are equivalent .
the intrinsic spin order parameter @xmath52 in the ground state can be determined by identifying and calculating the respective correlation functions that reflect the emerging long - range order @xcite , @xmath53 for each phase @xmath54 , where @xmath55 is the average over the ground state @xmath56 . in this definition
we select @xmath57 sign for the spin components @xmath58 for the nel af phase and @xmath59 sign for the staggered and zz phase ; @xmath60 for the nel ( @xmath61 ) , @xmath62 for the st ( @xmath63 ) , and either @xmath64 or @xmath65 for the zz ( @xmath66 ) phase ( these points are equivalent in this case ) . here
@xmath21 and @xmath67 label unit cells [ fig .
[ fig : defs](a ) ] , and @xmath68 .
the order parameter @xmath52 ( [ ssf ] ) is large when spin correlations are close to the ones expected for a magnetic phase @xmath54 ; in all other phases it is negligible .
two examples are shown in fig .
[ fig : defsb](a ) : ( i ) for @xmath69 one finds large @xmath70 , and ( ii ) for @xmath71 this order parameter drops to @xmath72 but @xmath73 increases to @xmath74 .
indeed , a transition between the af and zz phase is found here , see dotted line in fig .
[ fig : defsb](b ) , while the st phase is unstable here and @xmath75 is small .
( [ ssf ] ) representing the nel ( @xmath54=af ) , the zigzag ( @xmath54=zz ) and the kitaev invariant @xmath76 ( [ w ] ) obtained for @xmath77 and @xmath78 .
( b ) phase diagram of the kh model eq . in @xmath79 plane ( points ) for @xmath77 , with af ,
st , zz and ksl phases the insets show types of magnetic order ( arrows ) or disorder ( circles ) .
, title="fig:",width=234 ] ( [ ssf ] ) representing the nel ( @xmath54=af ) , the zigzag ( @xmath54=zz ) and the kitaev invariant @xmath76 ( [ w ] ) obtained for @xmath77 and @xmath78 .
( b ) phase diagram of the kh model eq . in @xmath79 plane ( points ) for @xmath77 , with af , st , zz and ksl phases the insets show types of magnetic order ( arrows ) or disorder ( circles ) .
, title="fig:",width=264 ] + in the ksl phase spin correlations @xmath80 vanish beyond nn spins at @xmath81 @xcite , and further neighbor correlations remain small in the kitaev liquid regime at @xmath82 @xcite .
this results in @xmath83 for all conventional spin order parameters @xcite . to identify the ksl
we introduce here an average of the kitaev invariant @xcite on a single hexagon @xmath84 , @xmath85 where @xmath86 labels the spin component @xmath87 interacting with a spin at site @xmath67 along the outgoing bond @xmath88 via @xmath89 .
one finds that @xmath90 when the ground state of the kitaev model is approached at @xmath91 , see fig .
[ fig : defsb](a ) .
the phase diagram in the @xmath11 versus @xmath92 plane is displayed in fig .
[ fig : defsb](b ) .
we recognize the af phase at small @xmath92 , the intermediate st phase , and at large @xmath93 the kitaev liquid phase . at large further neighbor exchange interaction
@xmath11 the zigzag phase emerges in the intermediate range of @xmath92 .
theses phases were identified by the order parameters discussed above , while the phase boundaries were determined by a different powerful tool , namely the study of the fidelity susceptibility @xcite , i.e. , the changing rate of the overlap between ground states at adjacent points .
note that the af@xmath94zz transition at @xmath95 follows from symmetry arguments and as such is independent of the cluster size .
in the following we shall analyze the spectral properties of a hole inserted into the ordered ground state @xmath56 , being the quantum nel af , st , or zz phase .
the ksl phase will be explored in detail in the subsequent section .
we use here the standard numerical lanczos algorithm which spans efficiently the relevant krylov space and yields spectral functions in form of a continued fraction @xcite .
the calculation begins with the determination of the ground state @xmath56 and the subsequent addition of a hole , that is the annihilation of an electron as in a photoemission experiment .
therefore we consider in the following the hole creation operator , @xmath96 , in form of a plane wave that includes all sites of the honeycomb lattice equally , and alternatively a hole creation operator , @xmath97 , where holes are created only on one sublattice @xcite , namely sublattice @xmath42 , @xmath98 when a hole is created in the ground state @xmath56 , the spectral functions , @xmath99 correspond to the physical green s function @xmath100 that is measured in arpes experiments ( [ ac ] ) , or to the sublattice green s function @xmath101 ( [ ad ] ) . in the definition of the spectral functions
( [ ac ] ) and ( [ ad ] ) excitation energies are measured relative to the ground state energy @xmath102 of a mott insulator with @xmath103 electrons .
( [ ac ] ) for one hole excitations @xcite at strong coupling , @xmath104 , obtained by ed at four distinct momenta @xmath105 ( solid , dashed , dashed - dotted and dotted lines ) for : ( a ) the nel phase at @xmath106 , ( b , c ) st and zz phases at @xmath107 , and ( d ) the ksl phase at @xmath108 .
parameters : @xmath77 , @xmath109 , and @xmath110 , except in ( c ) where @xmath78.,title="fig:",width=291 ] + in all phases the spectra @xmath111 and @xmath112 shown in figs .
[ fig : acw ] and [ fig : adw ] , respectively , have the total width @xmath113 as for free hole motion on the honeycomb lattice . for small @xmath114
the model is weakly frustrated and @xmath56 is the quantum af nel state , see fig .
[ fig : defsb](b ) . taking @xmath115 , a value representative for strong coupling ( @xmath116 ) @xcite , i.e.
, when the kinetic energy of a hole is larger than the energy of a magnetic bond , one finds that the spectral function @xmath111 has a qp at low energy and its spectral weight is large at the @xmath117 and much weaker at the @xmath3 point , see fig .
[ fig : acw](a ) . in this phase
no qualitative differences between @xmath111 and @xmath112 functions are observed , see figs .
[ fig : acw](a ) and [ fig : adw](a ) , except near @xmath118 for the @xmath3 point .
the sublattice spectral function , @xmath112 , where holes are injected / removed on the same sublattice , reveals large spectral weight at low energy . for the st and zz phases the low energy features in @xmath112
can indeed be identified as qps accompanied by incoherent spectral weight they are more pronounced in the zz phase , cf . figs .
[ fig : adw](b ) and [ fig : adw](c ) .
yet , the qp features are suppressed at the @xmath3 and @xmath119 points in these phases in the physical spectral function @xmath111 , cf . figs . [
fig : acw](b ) and [ fig : acw](c ) . as shown before @xcite ,
this is consistent with the observed absence of qps in arpes for na@xmath4iro@xmath5 @xcite .
( [ ad ] ) for a hole excitation created on one sublattice @xcite .
the method , lines and parameters are the same as in fig .
[ fig : acw ] . , title="fig:",width=291 ] + the momentum @xmath105 dependence of the low energy qps shown in fig .
[ fig : disp ] reveals the strong dependence of hole dispersion on the magnetic order . as in cuprates
@xcite , the qp dispersion in the nel af phase is narrowed from the free ( unconstrained ) fermionic band width @xmath120 by strong correlations and is determined by the magnetic exchange @xmath121 .
the dispersion has a minimum ( maximum ) at the @xmath122 ( @xmath3 ) point and is further reduced when frustration of magnetic exchange increases from @xmath123 to @xmath106 [ fig .
[ fig : disp](a ) ] . at finite @xmath11 the hole energy decreases at the @xmath119 point , but otherwise the dispersions at @xmath106 are quite similar , cf . figs .
[ fig : disp](a ) and [ fig : disp](b ) . ) by ed for @xmath115 , @xmath77 , and increasing @xmath92 : ( a ) @xmath110 , and ( b ) @xmath124 .
hole energies are respective to the ground state energy @xmath102.,title="fig:",width=302,height=226 ] ) by ed for @xmath115 , @xmath77 , and increasing @xmath92 : ( a ) @xmath110 , and ( b ) @xmath124 .
hole energies are respective to the ground state energy @xmath102.,title="fig:",width=302,height=226 ] + in contrast , the dispersion is absent in the st phase , see fig . [
fig : disp](a ) , as coherent hole propagation is hindered here due to the alternating af and fm bonds . instead
, for fm chains in the zz phase the dispersion appears reversed with respect to that found for the nel phase now a minimum ( maximum ) is at the @xmath3 ( @xmath122 ) point . while this dispersion decreases from weak ( @xmath125 ) to strong ( @xmath126 ) coupling
, its shape remains the same [ dashed line in fig .
[ fig : disp](b ) ] . at first glance
one might conclude from the spectral functions for the ksl , displayed in figs .
[ fig : acw](d ) and [ fig : adw](d ) , that hole propagation in the ksl is similar to that in the zz phase [ figs .
[ fig : disp](a ) and [ fig : disp](b ) ] . in this case
the lowest excitation energy is found at the @xmath3 point .
again , one finds a distinct peak in @xmath127 which is absent in @xmath128 , suggesting also here a hidden qp . as shall see below that the fine structure of the low energy peaks in the spectral function shown in fig . [
fig : adw](d ) does not represent well defined qps in the case of the ksl .
next we show that the ksl phase is manifestly different from all the ordered phases discussed so far .
we address the nature of low energy states by analyzing first the intermediate coupling regime of @xmath129 .
again one finds distinct low energy peaks in @xmath127 @xcite , see fig .
[ fig : ksl](b ) , missing in @xmath130 [ fig .
[ fig : ksl](a ) ] .
in contrast , both spectral functions are rigorously identical at the @xmath122 point , cf . @xmath131 and @xmath132 in figs . [
fig : ksl](a ) and [ fig : ksl](b ) .
we note that for the honeycomb lattice @xmath122 corresponds to the dirac point which is the degeneracy point of noninteracting electrons @xcite .
a surprise comes when the fine structure of the sublattice spectral function is analyzed in absence of spectral broadening [ at @xmath133 in eq .
( [ ad ] ) ] , @xmath134 which may be rewritten using spectral weights : @xmath135 in the following we shall see the advantage of studying the single - particle propagation directly in terms of the _ spectral weight distribution _ , @xmath136 @xmath137 and @xmath138 is an excited state in the space with one extra hole and total momentum @xmath139 that contributes with a finite spectral weight .
we stress , that @xmath140 is in mathematical terms a distribution and not a function , although in the tl it may contain parts that can be represented by continuous curves , in some cases . when looking at the spectral weight distribution in fig .
[ fig : ksl](c ) we recognize a robust qp , that is , a well separated bound state at lowest energy , only at the @xmath3 point with large spectral weight @xmath141 . at _ all other _
@xmath142 points no bound states exist but instead rather continuous distributions of weights @xmath140 is found in the range of the lowest eigenvalues @xmath143 ( [ omega ] ) . from the spectral weight distribution displayed in fig .
[ fig : ksl](c ) it is clear that for example at the @xmath119 point the spectra are represented by a superposition of several many - body states in the @xmath144-particle sector with slightly different energies and there is no dominant pole that could be considered as a qp state . on the other hand , simply by looking at the spectral function @xmath145 in fig .
[ fig : ksl](b ) one may easily overlook the breakdown of the qp picture .
we stress that in all cases there is substantial incoherent spectral weight at higher energies that extends to the upper edge of the spectrum at @xmath146 . to understand this striking difference of the character of the spectra at low energy at @xmath3 and @xmath147 , respectively , we need to investigate the spin excitations in the kitaev liquid regime .
finally it is the scattering of carriers from spin excitations that determines whether they can propagate as qps or whether they are completely overdamped . at four nonequivalent
momenta @xmath142 ( solid , dashed , dashed - dotted and dotted lines ) with @xmath109 : ( a ) full spectral function @xmath111 , and ( b ) sublattice function @xmath112 .
panel ( c ) shows the spectral weights @xmath148 ( [ alpha ] ) ( symbols ) larger than @xmath149 at low energy ( shaded ) for different @xmath142 values ; dashed lines are guides to the eye .
parameters : @xmath108 , @xmath150 .
, title="fig:",width=302 ] + our aim here is to explore the spin excitations of the kh model in the ksl regime .
however , our discussion will be more transparent when we first focus on the pure kitaev model at @xmath81 and subsequently analyze by numerical simulation the changes of the spin structure factor for the kh case in the ksl regime ( @xmath151 ) .
we shall employ here the usual representation of the kitaev model in terms of @xmath152 spin operators @xmath153 where @xmath154 and @xmath81 , according to eq . .
spin excitations are most easily understood by transforming the kitaev model into the majorana representation @xcite .
kitaev introduced a representation for the spin algebra @xmath155=i \epsilon_{abc}\sigma_i^c\delta_{ij}$ ] in terms of four majorana operators @xmath156 , @xmath157 , per lattice site which obey the anticommutation relations , @xmath158 . here each spin operator component
is expressed by a product of two majorana fermions , @xmath159 where @xmath160 is associated with the bond direction and @xmath161 , with the respective vertex ( at site @xmath21 ) .
we suppress the upper index in @xmath162 further on . using this representation
one can write the hamiltonian as follows , @xmath163 where @xmath164 are the bond operators , @xmath165 to take care of the fermionic property @xmath166 we adopt a notation where @xmath167 is on the @xmath168 sublattice , respectively .
the important point is now , that the so - defined bond operators commute with the hamiltonian , @xmath169=0.\ ] ] hence they are conserved quantities and due to their definition as products of two majoranas , see eq . , these operators can only take the values @xmath170 . in the ground state sector
all @xmath164 have the same sign on all bonds , either @xmath171 or @xmath172 .
a change of a bond variable leads to a _ vortex pair excitation _ with a gap @xmath173 .
in addition there is a second class of spin excitations which result from the motion of the @xmath174 majorana particles described by the hamiltonian in eq . .
here we concentrate on the symmetric case , with equal exchange constants along three directions in the honeycomb lattice , @xmath175 ; in this case the _ majorana excitations are gapless _
@xcite . for the undoped ksl in the case of the kitaev - heisenberg model eq .
at four distinct momenta , for : ( a ) @xmath108 , and ( b ) @xmath176 .
the curves are obtained by exact diagonalization of a 24-site cluster with periodic boundary conditions and are smoothened using a lorentzian broadening @xmath177.,title="fig:",width=302 ] for the undoped ksl in the case of the kitaev - heisenberg model eq . at four distinct momenta , for : ( a ) @xmath108 , and ( b ) @xmath176 .
the curves are obtained by exact diagonalization of a 24-site cluster with periodic boundary conditions and are smoothened using a lorentzian broadening @xmath177.,title="fig:",width=313 ] next we shall investigate the dynamic spin - structure factor in the kitaev limit @xmath178 by exact diagonalization and explore the changes in the kitaev - liquid regime of the kh model for @xmath82 , which can not be solved analytically .
this is demonstrated by a calculation of the dynamic spin - structure factor @xmath179 defined as follows @xcite , @xmath180 , and @xmath181 denotes spin raising ( lowering ) operators , respectively .
here @xmath56 and @xmath182 are the ground and excited state , with energies @xmath183 and @xmath184 , respectively .
the spin quantization axis is chosen parallel to the @xmath51-th spin axis of the kitaev term for convenience .
figure [ fig : skw](a ) shows @xmath185 for the kh model eq .
( 1 ) in the ksl regime at @xmath108 for four different @xmath139 points : @xmath3 , @xmath122 , @xmath186 and @xmath119 .
the numerical calculations were performed for a 24-site cluster with periodic boundary conditions .
the main weight of the spin structure factor is concentrated between the vortex type spin gap at @xmath187 and @xmath188 .
we observe a moderate momentum dependence with the vortex spin gap at @xmath3 , @xmath189 slightly larger than at @xmath122 , @xmath186 , and @xmath119 , where we find @xmath190 .
moreover , we can conclude that the dispersion of the gap and of the peak structures is due to the heisenberg terms in the hamiltonian and disappears when one approaches the kitaev limit @xcite , as seen in fig .
[ fig : skw](b ) .
thus we find here that the main contributions to the spin - response as measured by dynamical spin - structure factor come from the gapped vortex excitations .
furthermore the classification of spin excitations obtained for the pure kitaev model as well as the spin gap due to vortex excitations appears still relevant for the ksl phase of the kh model eq .
( 1 ) at @xmath108 .
our central argument why qps at finite momentum are destroyed at _ intermediate coupling _ , as shown in fig .
[ fig : ksl](c ) , will be outlined next .
our explanation rests on the fact that there are two distinct types of elementary spin excitations from which the holes can scatter in the ksl .
these excitations can be classified according to the exact solution given by kitaev @xcite as : ( i ) gapped vortex spin excitations with minimal gap @xmath191 , and ( ii ) gapless majorana excitations .
we consider as intermediate coupling regime the range @xmath192 , where the kinetic energy of holes is comparable with the magnetic energy on the bonds , and not much larger as in the strong coupling regime , @xmath193 . an important aspect of the spectral functions at intermediate coupling , as in the case of fig .
[ fig : ksl](c ) which was calculated for @xmath194 , is the large size of the spin gap , @xmath195 in units of the @xmath0-scale . as the excitation energies @xmath143 of the low energy states of holes at @xmath119 , @xmath186 and @xmath122 relative to the bottom of the band at the @xmath3 point @xmath196 ,
are much smaller than the vortex spin gap @xmath197 at @xmath194 , no decay via vortex excitations @xcite can occur .
therefore , we conclude that the destruction of qps is due to the scattering of _ gapless _ majorana fermion excitations , and that fig .
[ fig : ksl](c ) has to be seen as a fingerprint of fractionalization of electrons into holons and gapless majorana fermions of the kitaev model @xcite . because of the dominant role of scattering from gapless spin - excitations we expect that in larger clusters with a denser @xmath139-mesh , hole pockets near the @xmath3 point are not protected against strong scattering , and in consequence the ksl does not turn into a fermi liquid at low doping .
, as obtained for : ( a ) the zz phase and ( b ) the ksl phase , both at intermediate coupling @xmath198 , and ( c ) the ksl phase at strong coupling @xmath126 .
parameters : ( a ) @xmath199 and @xmath200 ; ( b ) and ( c ) @xmath108 and @xmath110 in eq .
width=302 ] before moving to the strong coupling regime @xmath201 we shall highlight the striking difference of the spectral weight distribution @xmath202 at low energy in the zz and the ksl phase , respectively , at @xmath198 . in fig .
[ fig : qp](a ) the spectral weight distribution @xmath202 for the zz phase shows well defined qp bound states .
although the spectral weights of these qps are significantly reduced , the bound states are well separated from the continuum ( of incoherent states ) at higher energy , except for @xmath203 where this binding energy is much weaker . in the ksl phase fig . [
fig : qp](b ) , on the other hand , a well defined bound state is seen only at the @xmath3 point . because of the size of the vortex gap @xmath204 the absence of qps in fig .
[ fig : qp](b ) at @xmath119 , @xmath186 and @xmath122 is a smoking gun for the important role played by scattering by gapless majorana excitations .
this is consistent with the spectral shape of @xmath205 which is reminiscent of a continuum , that may result from a convolution of holons and gapless majorana fermions .
we turn now to the _ strong coupling _ regime . in the strong coupling case of @xmath115 ,
i.e. , relevant for the iridates @xcite and displayed in fig .
[ fig : qp](c ) , the vortex spin gap becomes small ( in units of @xmath0 ) , i.e. , @xmath206 t .
thus the strong coupling result in fig .
[ fig : qp](c ) highlights the effect of the additional vortex excitations which form a new decay channel and damp the excitations at the @xmath122 , @xmath186 , and @xmath119 points even further .
compared to the result at intermediate coupling at @xmath207 [ fig .
[ fig : qp](b ) ] where the vortex gap is @xmath208 , at strong coupling one finds a spin gap to the vortex excitations which is 20 times smaller , that is @xmath209 .
furthermore , it may be instructive to go back to the spectral function @xmath210 in fig . [
fig : adw](d ) which basically contains the same information for the strong coupling case ( @xmath126 ) of the ksl as the spectral weight distribution shown in fig .
[ fig : qp](c ) . the fine structure seen in @xmath202
can only be resolved in @xmath210 when the resolution parameter @xmath174 is taken small enough . for arpes experiments , where actually @xmath211 is measured
, this implies that a sufficient momentum and energy resolution is required .
we conclude , that the peak at @xmath3 , which appeared as a separate bound state at intermediate coupling in fig .
[ fig : qp](b ) , appears now at strong coupling , see fig .
[ fig : qp](c ) , rather as the edge of the continuum than as an isolated bound state .
moreover , when a single hole does not propagate as a qp , then a dilute gas of holes will not form a fermi liquid .
therefore we stress , that the absence of a well defined separate bound state is a _ new _ qualitative feature which speaks against qps and fermi liquid behavior at low doping , and this time the argument emerges from a diagnosis at the @xmath3 point itself !
we have shown that hole propagation is modified in a remarkable way as increasing kitaev interactions drive the system from the nel order via other ordered antiferromagnetic phases towards the kitaev spin liquid .
quasiparticles are found in the nel phase , whereas coherent hole propagation is hindered in stripe and zigzag phases , where hidden quasiparticles with weak dispersion result from coexisting ferromagnetic and antiferromagnetic bonds . as the most unexpected result , in the kitaev liquid phase we have found unprecedented spectral weight distribution at low energy that signals the absence of quasiparticles , both at intermediate and strong coupling .
thus , it appears clearly that carrier motion in the lightly doped kitaev liquid is non - fermi liquid like .
the above conclusion follows from the short - range nature of spin correlations in the kitaev spin liquid @xcite . unlike in a quantum antiferromagnet on a square @xcite or honeycomb @xcite lattice
where spin - flip processes couple to a moving hole and generate new energy scale for coherent hole propagation , the kitaev spin - liquid phase is characterized by ising - like nearest neighbor spin correlations of one spin component .
such correlations are insufficient to generate coherent hole propagation .
they are well captured by the present cluster size of @xmath32 sites , and qualitative changes of the described scenario are therefore unexpected when the cluster size is increased . summarizing , we have given clear arguments that shed serious doubts on the claim from a slave - boson approximation , namely that the low doped kitaev spin - liquid phase is a fermi liquid @xcite .
the first of these arguments rests on the exact solution for the spin excitations in the kitaev model in the intermediate coupling regime , and on our finding that _ gapless _ majorana excitations are responsible for the absence of qps away from the @xmath3 point . from the gaplessness
we conclude that also states in the closer vicinity of @xmath3 are not protected .
our second argument emerges in the strong coupling case from the result for the spectral distribution @xmath212 at the @xmath3 point itself , which in this case has no similarity to a qp but appears rather as the edge of a continuum .
we consider this as evidence that at strong coupling there are no qps in the single hole case near the @xmath3 point , and from this finding we can safely conclude that fermi liquid behavior is absent in the low doping regime .
we thank bruce normand for valuable advice , as well as mona berciu , george jackeli and roser valent for insightful discussions .
a.m.o . acknowledges support by the polish national science center ( ncn ) under project no .
2012/04/a / st3/00331 .
acknowledges support by the natural science foundation of jiangsu province of china under grant no .
bk20141190 .
s. a. trugman , phys .
b * 37 * , 1597 ( 1988 ) ; c. d. batista and g. ortiz , phys . rev
* 85 * , 4755 ( 2000 ) ; m. daghofer , k. wohlfeld , a. m. ole , e. arrigoni , and p. horsch , _ ibid . _ * 100 * , 066403 ( 2008 ) ; k. wohlfeld , m. daghofer , a. m. ole , and p. horsch , phys .
b * 78 * , 214423 ( 2008 ) ; k. wohlfeld , a. m. ole , and p. horsch , _ ibid . _ * 79 * , 224433 ( 2009 ) ; w. brzezicki , m. daghofer , and a. m. ole , _ ibid .
_ * 89 * , 024417 ( 2014 ) ; a. m. ole , j. phys . : condens .
matter * 24 * , 313201 ( 2012 ) . c. l. kane , p. a. lee , and n. read , phys .
b * 39 * , 6880 ( 1989 ) ; g. martinez and p. horsch , _ ibid .
_ * 44 * , 317 ( 1991 ) ; a. ramak and p. horsch , _ ibid . _ * 57 * , 4308 ( 1998 ) ; z. p. liu and e. manousakis , _ ibid . _ * 44 * , 2414 ( 1991 ) ; * 45 * , 2425 ( 1992 ) ; p. wrbel and r. eder , _ ibid . _ * 58 * , 15160 ( 1998 ) ; j. bona , s. maekawa , and t. tohyama , _ ibid . _
* 76 * , 035121 ( 2007 ) ; i. j. hamad , a. e. trumper , a. e. feiguin , and l. o. manuel , _ ibid . _
* 77 * , 014410 ( 2008 ) .
jiang , z .- c .
qi , and s. trebst , phys .
b * 83 * , 245104 ( 2011 ) ; j. reuther , r. thomale , and s. trebst , _ ibid .
_ * 84 * , 100406 ( 2011 ) ; r. schaffer , s. bhattacharjee , and y. b. kim , _ ibid . _ * 86 * , 224417 ( 2012 ) ; i. kimchi and y. z. you , _ ibid .
_ * 84 * , 180407 ( 2011 ) ; y. yu , l. liang , q. niu , and s. qin , _ ibid . _
* 87 * , 041107 ( 2013 ) .
y. singh and p. gegenwart , phys .
b * 82 * , 064412 ( 2010 ) ; x. liu , t. berlijn , w .- g .
yin , w. ku , a. tsvelik , y .- j .
kim , h. gretarsson , y. singh , p. gegenwart , and j. p. hill , _ ibid . _ * 83 * , 220403 ( 2011 ) .
a. f. albuquerque , d. schwandt , b. hetnyi , s. capponi , m. mambrini , and a.m. lachli , phys .
b * 84 * , 024406 ( 2011 ) ; d. c. cabra , c. a. lamas , and h. d. rosales , _ ibid . _
* 83 * , 094506 ( 2011 ) ; h. d. rosales , d. c. cabra , c. a. lamas , p. pujol , and m.e .
zhitomirsky , _ ibid . _ * 87 * , 104402 ( 2013 ) ; a. kalz , m. arlego , d. cabra , a. honecker , and g. rossini , _ ibid .
_ * 85 * , 104505 ( 2012 ) .
r. ganesh , j. van den brink , and s. nishimoto , phys .
* 110 * , 127203 ( 2013 ) ; r. ganesh , s. nishimoto , and j. van den brink , phys . rev .
b * 87 * , 054413 ( 2013 ) ; z. zhu , d.a . huse , and s. r. white , phys . rev .
* 110 * , 127205 ( 2013 ) . | we explore with exact diagonalization the propagation of a single hole in four magnetic phases of the @xmath0-@xmath1-like kitaev - heisenberg model on a honeycomb lattice : the nel antiferromagnetic , stripe , zigzag and kitaev spin - liquid phase .
we find coherent propagation of spin - polaron quasiparticles in the antiferromagnetic phase by a similar mechanism as in the @xmath0-@xmath1 model for high-@xmath2 cuprates . in the stripe and zigzag phases clear quasiparticles features appear in spectral functions of those propagators where holes are created and annihilated on one sublattice , while they remain largely _ hidden _ in those spectral functions that correspond to photoemission experiments . as the most surprising result , we find a totally incoherent spectral weight distribution for the spectral function of a hole moving in the kitaev spin - liquid phase in the strong coupling regime relevant for iridates . at intermediate coupling
the finite systems calculation reveals a well defined quasiparticle at the @xmath3 point , however , we find that the gapless spin excitations wipe out quasiparticles at finite momenta . also for this more subtle case we conclude that in the thermodynamic limit the lightly doped kitaev liquid phase does not support quasiparticle states in the neighborhood of @xmath3 , and therefore yields a _ non - fermi liquid _ , contrary to earlier suggestions based on slave - boson studies .
these observations are supported by the presented study of the dynamic spin - structure factor for the kitaev spin liquid regime . |
@xmath2 car is a very massive ( @xmath3 solar masses ) star known for its strong mass outflow eruptions , and is one of the most interesting objects of our galaxy ( e.g. , davidson & humphrey 1997 ) .
radial velocity variations of spectroscopic lines accumulated over the years provide a strong evidence that is a binary system ( e.g. , damineli et al .
2008a , b ) where the primary star is a luminous blue variable ( lbv ) star orbiting in a very eccentric binary ( @xmath4 ) with a companion star believed to be an o star of @xmath5 solar masses .
the orbital period is 5.53 years ( @xmath62023 days ) ( e.g. , damineli et al . , 2008a ) : the system has been monitored in the radio , mm , ir , optical and x - ray bands for at least three cycles .
both stars emit dense and high - velocity gaseous winds , and the binary system is ideal to study the interaction of colliding winds and to test theories of particle acceleration and radiation under extreme conditions .
the mass outflow rates and wind speeds of the two stars inferred from the wealth of all available data are @xmath7 @xcite .
eta car is then interesting among other colliding wind binaries since the observable x - ray emission in the 2 - 10 kev band is almost entirely produced by the shocked fast wind of the secondary star , with little if any contribution from the slow shocked wind of eta car itself .
the system is known for its variability and occasional erratic eruptions detected in the ir and optical bands , as well as for its distinct asymmetric pattern of optical line and x - ray emission during its orbital period @xcite . has been repeatedly observed in the energy ranges 1 - 10 kev and 20 - 100 kev by different observatories .
it is certainly the only source showing a non - thermal x - ray spectrum within a region centered on with a 1 degree diameter ( the anomalous x - ray pulsar axp 1e 1048.1 - 5937 is about 0.6 degrees away ) . whereas the 1 - 10 kev spectrum is dominated by a quasi - thermal and variable component @xcite
, the hard x - ray observations show non - thermal emission that appears to vary along the orbit @xcite .
the source was detected with high significance by bsax - pds and integral - isgri far from periastron .
integral is capable of resolving field sources with a few arcminute resolution in the hard x - ray range .
although integral observed the system at different phase periods ( 0.99 - 0.01 , 0.160.19 , 0.35 - 0.37 ) , was detected only during the phase interval 0.160.19 with an average 22 - 100 kev x - ray flux of @xmath8 @xcite .
the carina region has been observed at gamma - ray energies above a few mev by the osse , comptel and egret instruments on board of the compton gamma - ray observatory ( cgro ) .
an egret gamma - ray source ( 3eg j1048 - 5840 ) is catalogued at about 1 degree distance from . however , no gamma - ray emission above 100 mev has been reported by cgro from the region . the gamma - ray astrophysics mission agile @xcite observed several times the carina region in the galactic plane during its early operational phases and cycle-1 observations . here
we report the main results of the gamma - ray observations of the region carried out by the agile satellite during the period 2007 july - 2009 january , simultaneously in the energy bands 30 mev - 30 gev and 18 - 60 kev . a high - confidence gamma - ray source ( )
was detected positionally consistent with by integrating all data , as well as by considering specific observation periods .
the agile mission has been operating since 2007 april @xcite .
the agile scientific instrument is very compact and is characterized by two co - aligned imaging detectors operating in the energy ranges 30 mev - 30 gev ( grid , barbiellini et al .
2002 , prest et al . , 2003 ) and 18 - 60 kev ( super - agile , feroci et al . 2007 ) , as well as by an anticoincidence system @xcite and a calorimeter @xcite .
agile s performance is characterized by large fields of view ( 2.5 and 1 sr for the gamma - ray and hard x - ray bands , respectively ) , optimal angular resolution and good sensitivity ( see tavani et al .
2008 for details about the mission and main instrument performance ) .
flux sensitivity for a typical 1-week observing period can reach the level of several tens of @xmath9 above 100 mev , and 10 - 20 mcrab in the 18 - 60 kev range depending on off - axis angles and pointing directions . [ cols="^,^,^,^,^,^,^,^ " , ] [ tab-1 ] ( * ) average orbital phase of calculated at the center of the time interval .
+ ( * * ) square root of the the maximum likelihood test statistic ( ts ) representing the statistical significance of the detection .
+ ( * * * ) gamma - ray flux of above 100 mev in units of @xmath10 s@xmath11 obtained by taking into account the nearby source in the multisource likelihood analysis .
we also indicate 2-sigma upper limits in the same units .
+ ( * * * * ) during this period the source reached the gamma - ray flux above 100 mev of @xmath12 .
the agile satellite repeatedly pointed at the carina region for a total of @xmath13 days during the time period 2007 july - 2009 january .
table 1 summarizes the agile observations of the field .
analysis of the gamma - ray data was carried out with the @xmath14 calibrated filter , with a gamma - ray event selection that takes into account standard saa event cuts and 80 degree earth albedo filtering .
we used the agile gamma - ray software release build 17 and the standard hard x - ray analysis software .
car region above 100 mev summing all data collected from 2007 july to 2008 october .
the central gamma - ray source that can be associated with @xmath2 car is ; we also indicate the prominent nearby gamma - ray source which is associated with the radio pulsar psr b1046 - 58 ( kaspi et al .
2006 ; abdo et al .
the color bar scale is in units of @xmath15 .
pixel size is 0.1 degrees , and we used a 3-bin gaussian smoothing .
white contour levels of the agile sources start from 0.0005 and increase in steps of 0.000028 .
the optical position of is marked by a small black circle .
the integral sources ( leyder et al . 2008 ) are marked with cyan circles .
, height=245 ] fig .
[ etacar - fig-3 ] shows the integrated sky map of the region at energies above 100 mev for the period 2007 july - 2008 october .
a gamma - ray source is detected with high confidence ( 7.8 sigma ) at the position @xmath16 . the average gamma - ray flux above 100 mev and integrated over the whole period 2007 july - 2008 october is @xmath17 .
we call this source following the source designation of the first agile catalog of high - confidence gamma - ray sources ( pittori et al . 2009 ) .
we included in all our multisource analysis the nearby gamma - ray source that the agile - grid detects with an average and constant gamma - ray flux above 100 mev of @xmath18 .
is well within the 95% confidence radius gamma - ray error box of ; the other nearby hard x - ray sources in the field ( the anomalous x - ray pulsar axp 1e 1048.1 - 5937 , and igr j10447 - 6027 ) are excluded .
the super - agile hard x - ray imager did not detect a source coincident with for both short and long integrations of table 1 .
depending on the source position in the fov , the typical 3-sigma super - agile upper limit is 10 - 20 mcrab , i.e. , consistent with the integral detection and upper limits of .
we searched for short timescale variability of the gamma - ray and hard x - ray flux from throughout the whole agile observing periods of the carina region .
a 2-day gamma - ray flare from the direction of was detected during the observation of 2008 october 10 - 17 .
the emission reached its peak gamma - ray emission during the period 2008 oct .
11 ( 02:57 ut ) - 2008 oct . 13
( 04:16 ut ) .
our analysis gives a 5.2 sigma detection of a source at the position @xmath19 fully consistent with the position but with a gamma - ray flux above 100 mev of @xmath12 .
[ etacar - fig-4 ] shows the time sequence of 2-day integration gamma - ray maps of the region during the period 2008 10 - 17 october .
[ etacar - fig-5 ] shows the agile gamma - ray data of table 1 superimposed with the simultaneous rxte ( pcu2 net rate ) lightcurve in the energy band 2 - 15 kev during the period 2007 february - 2009 january ( the typical and relatively abrupt decrease of the x - ray emission near periastron is clearly visible ) .
[ etacar - fig - spectrum ] shows two representative different broad - band spectral states of during the period 2007 july 2008 october together with the historical x - ray and hard x - ray data reported from .
we mark in blue the average spectrum obtained by integrating all data outside periastron , and with a red cross the flux corresponding to the flaring state of 2008 october 11 - 13 .
we also report in the same plot the ( non simultaneous ) bsax - mecs and the integral - isgri spectral states of reported in the literature @xcite .
it is interesting to note that if is associated with , the average agile spectrum together with the integral historical spectrum is in qualitative agreement with expectations based on inverse compton and/or pion decay models of gamma - ray emission from colliding wind binaries ( e.g. , reimer , pohl , reimer 2006 ) . for the 2008 11 - 13 october flaring episode , the super - agile 18 - 60
kev upper limit is 70 mcrab in the energy band 18 - 60 kev .
obtaining simultaneous hard x - ray and gamma - ray data during the flaring state of is crucial to study the broad - band variability of the source .
however , due to unfavorable source positioning of in the super - a field of view in mid - october , 2008 , the hard x - ray upper limit is not very constraining .
agile pointed at the carina region during the period 2009 12 - 19 january as a special repointing to cover the periastron passage ( calculated to be occurring on 2009 january 11 ) .
we can not exclude , at this stage , the existence of a weak gamma - ray source consistent with .
however , we can currently provide a 2-sigma upper limit to the emission above 100 mev of @xmath20 .
a more detailed analysis of the agile and multifrequency data during the periastron passage will be discussed elsewhere .
is located in the carina nebula that extends for several degrees in a galactic region characterized by dense molecular clouds , young stars and star formation sites .
however , within the error box itself is by far the strongest and hardest source in the 2 - 10 kev and 22 - 100 kev ranges .
another source inside the agile error box and 7 arcmin away from is the x ray binary hd 93162/wr 25 ( wn6+o4 , with an orbital period of 208 days , gamen et al .
this system is known to be a colliding wind system .
pollock & corcoran ( 2006 ) found significant variability , possibly periodic , of its x ray flux .
however , wr 25 was not detected in the hard x - ray range by integral during any of its observations .
furthermore , no other prominent hard x - ray source is known in the error box except for itself .
the two nearby hard x - ray sources detected by integral @xcite are outside the 95% confidence level error box of agile .
multiple gamma - ray sources within the error box can not be excluded but are unlikely . based on the accumulated multifrequency evidence and the nature of the source
, we consider the association of and as very likely .
we briefly elaborate below on the theoretical implications of our results assuming that is indeed the gamma - ray counterpart of . colliding wind binaries ( cwbs )
are ideal systems to test theories of hydrodynamical shocks and particle acceleration under extreme radiative conditions provided by the proximity of the two stars .
in particular , supersonic winds can form efficient shocks where electrons and protons can be accelerated through first - order fermi ( eichler & usov 1993 ) or other acceleration mechanisms .
inverse compton ( ic ) scattering of shock - accelerated particles in the presence of the very intense ir - optical - uv background of the nearby very bright stars provides a crucial ingredient in cwbs .
in addition to synchrotron and bremsstrahlung electron emissions , the ic emission can dominate the high energy spectrum at energies larger than several tens of kev up to mev - gev energies .
furthermore , if protons are efficiently accelerated , they can interact with the dense stellar outflows and produce gamma - rays by pion production and neutral pion decay ( e.g. , eichler & usov 1993 , benaglia & romero 2003 ; reimer et al .
all these ingredients are important for the system and detailed hydrodynamical modelling of the mass outflow have been developed @xcite .
a comprehensive and detailed theoretical analysis of our data is beyond the scope of this paper .
we outline here a few important points .
if is the gamma - ray counterpart , our data show the first remarkable detection of a colliding wind system at hundreds of mev energies , confirming the efficient particle acceleration and the highly non - thermal nature of the strong shock in a cwb . the average gamma - ray flux of translates into gamma - ray luminosity of @xmath21 for an distance of 2.3 kpc , corresponding to a fraction of a percent of the total wind kinetic power .
the 2008 oct .
11 - 13 flare episode has a luminosity of @xmath22 .
the average broad - band gamma - ray spectrum determined by agile is in qualitative agreement with expectations of cwb spectra as calculated for dominant ic and neutral pion decay processes @xcite .
while the gamma - ray flux of is roughly constant during the time span covered by our observations , a significant variability was detected on a few day time - scale in october 2008 .
this episode indicates that the gamma - ray emission can be associated with intermittent strong shock acceleration episodes and/or magnetic field enhancements to be expected for a very variable and inhomogeneous mass outflow from the stars of the system .
in particular , we note that the strong gamma - ray flaring episode occurred a few months before periastron , when the efficiency of transforming a mass outflow enhancement into particle acceleration is expected to increase because of the closeness of the two stars .
provides then some crucial ingredients regarding the formation of high - energy emission in cwbs : ( 1 ) strong variability of the mass outflows ; ( 2 ) a high - speed wind from the less massive companion ; ( 3 ) a radiative environment with a specific bath of soft photons from both stars ( ir , optical and uv fluxes ) that can illuminate the shock region and provide a time variable environment for enhanced ic emission in the 100 mev range and beyond .
the theoretical implications are far reaching .
the system would provide the first cwb to test the particle acceleration models for non - relativistic mass outflows under a specific set of physical conditions .
it is very important to assess the efficiency of the particle acceleration process in such a radiative environment .
a gamma - ray flaring episode lasting @xmath23 days implies a fast acceleration timescale and subsequent radiation and decay of the strong shock properties leading to the efficient emission .
if the gamma - ray emission is associated with our observations provide important data to test shock acceleration models .
future gamma - ray observations and analysis will further contribute to enlighten the emission mechanism and the ultimate origin of . | we present the results of extensive observations by the gamma - ray agile satellite of the galactic region hosting the carina nebula and the remarkable colliding wind binary eta carinae ( ) during the period 2007 july 2009 january .
we detect a gamma - ray source ( ) consistent with the position of .
if is associated with the car system our data provide the long sought first detection above 100 mev of a colliding wind binary .
the average gamma - ray flux above 100 mev and integrated over the pre - periastron period 2007 july 2008 october is @xmath0 corresponding to an average gamma - ray luminosity of @xmath1 for a distance of 2.3 kpc .
we also report a 2-day gamma - ray flaring episode of on 2008 oct .
11 - 13 possibly related to a transient acceleration and radiation episode of the strongly variable shock in the system . |
the success of sat technology in practical applications is largely driven by _ incremental solving_. sat solvers based on conflict - driven clause learning ( cdcl ) @xcite gather information about a formula in terms of learned clauses .
when solving a sequence of closely related formulae , it is beneficial to keep clauses learned from one formula in the course of solving the next formulae in the sequence .
the logic of quantified boolean formulae ( qbf ) extends propositional logic by universal and existential quantification of variables .
qbf potentially allows for more succinct encodings of pspace - complete problems than sat .
motivated by the success of incremental sat solving , we consider the problem of incrementally solving a sequence of syntactically related qbfs in prenex conjunctive normal form ( pcnf ) . building on search - based qbf solving with clause and cube learning ( qcdcl ) @xcite
, we present an approach to incremental qbf solving , which we implemented in our solver .
different from many incremental sat and qbf @xcite solvers , allows to add clauses to and delete clauses from the input pcnf in a stack - based way by _
push _ and _ pop _ operations . a related stack - based framework was implemented in the sat solver @xcite . a solver api with _
push _ and _ pop _ increases the usability from the perspective of a user .
moreover , we present an optimization based on this stack - based framework which reduces the size of the learned clauses .
incremental qbf solving was introduced for qbf - based bounded model checking ( bmc ) of partial designs @xcite .
this approach , like ours , relies on selector variables and assumptions to support the deletion of clauses from the current input pcnf @xcite .
the quantifier prefixes of the incrementally solved pcnfs resulting from the bmc encodings are modified only at the left or right end .
in contrast to that , we consider incremental solving of _ arbitrary _ sequences of pcnfs . for the soundness
it is crucial to determine which of the learned clauses and cubes can be kept across different runs of an incremental qbf solver .
we aim at a general presentation of incremental qbf solving and illustrate problems related to clause and cube learning .
our approach is _ application - independent _ and applicable to qbf encodings of _ arbitrary _ problems .
we report on experiments with constructed benchmarks .
in addition to experiments with qbf - based conformant planning using @xcite , our results illustrate the potential benefits of incremental qbf solving in application domains like synthesis @xcite , formal verification @xcite , testing @xcite , planning @xcite , and model enumeration @xcite , for example .
we introduce terminology related to qbf and search - based qbf solving necessary to present a general view on incremental solving . for a propositional variable @xmath0 , @xmath1 or @xmath2
is a _ literal _ , where @xmath3 denotes the variable of @xmath4 .
a _ clause _ ( _ cube _ ) is a disjunction ( conjunction ) of literals .
a _ constraint _ is a clause or a cube .
the empty constraint @xmath5 does not contain any literals . a clause ( cube ) @xmath6 is _ tautological _ ( _ contradictory _ ) if @xmath7 and @xmath8 .
a propositional formula is in _ conjunctive ( disjunctive ) normal form _ if it consists of a conjunction ( disjunction ) of clauses ( cubes ) , called cnf ( dnf ) . for simplicity ,
we regard cnfs and dnfs as sets of clauses and cubes , respectively . a quantified boolean formula ( qbf ) @xmath9 is in _ prenex cnf ( pcnf ) _ if it consists of a quantifier - free cnf @xmath10 and a _ quantifier prefix _ @xmath11 with @xmath12 where @xmath13 are _ quantifiers _ and @xmath14 are _ blocks _
( i.e. sets ) of variables such that @xmath15 and @xmath16 for @xmath17 , and @xmath18 .
the blocks in the quantifier prefix are _ linearly ordered _ such that @xmath19 if @xmath20 . the linear ordering is extended to variables and literals : @xmath21 if @xmath22 , @xmath23 and @xmath24 , and @xmath25 if @xmath26 for literals @xmath4 and @xmath27 .
we consider only _ closed _ pcnfs , where every variable which occurs in the cnf is quantified in the prefix , and vice versa .
a variable @xmath28 is _ universal _ , written as @xmath29 , if @xmath30 and _ existential _ , written as @xmath31 , if @xmath32 .
a literal @xmath4 is universal if @xmath33 and existential if @xmath34 , written as @xmath35 and @xmath36 , respectively .
an _ assignment _ is a mapping from variables to the truth values _ true _ and _ false_. an assignment @xmath37 is represented as a set of literals @xmath38 such that , for @xmath39 , if @xmath40 is assigned to false ( true ) then @xmath41 ( @xmath42 ) .
pcnf @xmath43 under an assignment @xmath37 _ is denoted by @xmath44 $ ] and is obtained from @xmath43 as follows : for @xmath39 , if @xmath42 ( @xmath45 ) then all occurrences of @xmath40 in @xmath43 are replaced by the syntactic truth constant @xmath46 ( @xmath47 ) , respectively .
all constants are eliminated from @xmath44 $ ] by the usual simplifications of boolean algebra and superfluous quantifiers and blocks are deleted from the quantifier prefix of @xmath44 $ ] . given a cube @xmath6 and a pcnf @xmath43 , @xmath48 : = \psi[a]$ ] is the formula obtained from @xmath43 under the assignment @xmath49 defined by the literals in @xmath6 .
the _ semantics _ of closed pcnfs is defined recursively .
the qbf @xmath46 is satisfiable and the qbf @xmath47 is unsatisfiable .
the qbf @xmath50 is satisfiable if @xmath51 $ ] and @xmath52 $ ] are satisfiable .
the qbf @xmath53 is satisfiable if @xmath51 $ ] or @xmath52 $ ] are satisfiable .
a pcnf @xmath43 is _ satisfied under an assignment _
@xmath37 if @xmath44 = \top$ ] and _ falsified _ under @xmath37 if @xmath44 = \bot$ ] .
satisfied and falsified clauses are defined analogously .
given a constraint @xmath6 , @xmath54 for @xmath55 denotes the set of universal and existential literals in @xmath6 .
for a clause @xmath6 , _ universal reduction _ produces the clause @xmath56 .
_ q - resolution _ of clauses is a combination of resolution for propositional logic and universal reduction @xcite . given two non - tautological clauses @xmath57 and @xmath58 and a pivot variable @xmath59 such that @xmath60 and @xmath61 and @xmath62 .
let @xmath63 be the _ tentative q - resolvent _ of @xmath57 and @xmath58 .
if @xmath64 is non - tautological then it is the _ q - resolvent _ of @xmath57 and @xmath58 and we write @xmath65 .
otherwise , @xmath57 and @xmath58 do not have a q - resolvent .
given a pcnf @xmath9 , a _
q - resolution derivation _ of a clause @xmath6 from @xmath43 is the successive application of q - resolution and universal reduction to clauses in @xmath43 and previously derived clauses resulting in @xmath6 .
we represent a derivation as a directed acyclic graph ( dag ) with edges ( 1 ) @xmath66 if @xmath67 and ( 2 ) @xmath68 and @xmath69 if @xmath65 .
we write @xmath70 if there is a derivation of a clause @xmath6 from @xmath43 .
otherwise , we write @xmath71 .
q - resolution is a sound and refutationally - complete proof system for qbfs @xcite .
a _ q - resolution proof _ of an unsatisfiable pcnf @xmath43 is a q - resolution derivation of the empty clause .
we briefly describe search - based qbf solving with conflict - driven clause learning and solution - driven cube learning ( qcdcl ) @xcite and related properties . in the context of _ incremental _ qbf solving , clause and cube learning requires a special treatment , which we address in section [ sec_inc_satisfiability ] .
given a pcnf @xmath43 , a qcdcl - based qbf solver successively assigns the variables to generate an assignment @xmath37 .
if @xmath43 is falsified under @xmath37 , i.e. @xmath44 = \bot$ ] , then a new learned clause @xmath6 is derived by q - resolution and added to @xmath43 .
if @xmath43 is unsatisfiable , then finally the empty clause will be derived by clause learning .
if @xmath43 is satisfied under @xmath37 , i.e. @xmath44 = \top$ ] , then a new learned _ cube _ is constructed based on the following _ model generation rule _ , _ existential reduction _ and _ cube resolution_. [ def_model ] given a pcnf @xmath72 , an assignment @xmath37 such that @xmath44 = \top$ ] is a _ model of @xmath43 .
_ an _ initial cube _
@xmath73 is a conjunction over the literals of a model @xmath37 .
given a cube @xmath6 , _ existential reduction _ produces the reduced cube @xmath74 .
given two non - contradictory cubes @xmath57 and @xmath58 , _ cube resolution _ is defined analogously to q - resolution for clauses , except that existential reduction is applied and the pivot variable must be universal .
the cube resolvent of @xmath57 and @xmath58 ( if it exists ) is denoted by @xmath75 .
if @xmath43 is satisfiable , then finally the empty cube will be derived by cube learning ( theorem 5 in @xcite ) .
whereas in clause learning initially clauses of the input pcnf @xmath43 can be resolved , in cube learning first initial cubes have to be generated by the model generation rule , which can then be used to produce cube resolvents . similar to q - resolution derivations ( dags ) of clauses and q - resolution proofs
, we define _ cube resolution derivations _ of cubes and _ proofs of satisfiability_. _ clause derivation : _ _ cube derivation : _ .
the literals in the initial cubes @xmath76 and @xmath77 have been omitted in the figure to save space . ] .
the literals in the initial cubes @xmath76 and @xmath77 have been omitted in the figure to save space . ]
[ ex_running ] given the satisfiable pcnf @xmath78 , where @xmath79 with @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 , and @xmath85 .
figure [ fig_derivations ] shows the derivation of the clauses @xmath86 and @xmath87 by q - resolution and universal reduction . the assignment @xmath88 is a model of @xmath43 by definition [ def_model ] .
hence @xmath89 is an initial cube .
existential reduction of @xmath76 produces the cube @xmath90 .
similarly , @xmath91 is a model of @xmath43 and @xmath92 is an initial cube .
existential reduction of @xmath77 produces the cube @xmath93 .
the cube @xmath94 is obtained by resolving @xmath95 and @xmath96 . finally , existential reduction of @xmath97 produces the empty cube @xmath98 , which proves that the pcnf @xmath43 is satisfiable .
a qcdcl - based solver implicitly constructs derivation dags in constraint learning . however , typically only selected constraints of these derivations are kept as learned constraints in an _ augmented cnf _ @xcite . [ def_aug_cnf ]
let @xmath9 be a pcnf .
the _ augmented cnf ( acnf ) _ of @xmath43 has the form @xmath99 , where @xmath11 is the quantifier prefix , @xmath10 is the set of original clauses , @xmath100 is a cnf containing the learned clauses , and @xmath101 is a dnf containing the learned cubes obtained by clause and cube learning in qcdcl . given an acnf @xmath102 and an assignment @xmath37 , the notation @xmath103 $ ] is defined similarly to pcnfs .
analogously to clause derivations , we write @xmath104 if there is a derivation of a cube @xmath6 from the pcnf @xmath105 . during a run of a qcdcl - based solver the learned constraints can be derived from the current pcnf .
[ prop_lcl_lcu_properties ] let @xmath106 be the acnf obtained by qcdcl from a pcnf @xmath9 .
it holds that ( 1 ) @xmath107 and ( 2 ) @xmath108 .
proposition [ prop_lcl_lcu_properties ] follows from the correctness of constraint learning in _ non - incremental _ qcdcl .
that is , we assume that the pcnf @xmath43 is not modified over time
. however , as we point out below , in _ incremental _ qcdcl the constraints learned previously might no longer be derivable after the pcnf has been modified .
[ def_constraint_correct ] given the acnf @xmath106 of the pcnf @xmath9 , a clause @xmath109 ( cube @xmath110 ) is _ derivable _ with respect to @xmath43 if @xmath111 .
otherwise , if @xmath112 , then @xmath6 is _ non - derivable_. due to the correctness of model generation , existential / universal reduction , and resolution , constraints which are derivable from the pcnf @xmath43 can be added to the acnf @xmath102 of @xmath43 , which results in a satisfiability - equivalent ( @xmath113 ) formula . [ prop_learning_satequiv ] let @xmath106 be the acnf of the pcnf @xmath9 .
then ( 1 ) @xmath114 and ( 2 ) @xmath115 .
we define _ incremental qbf solving _ as the problem of solving a sequence of @xmath116 using a qcdcl - based solver .
thereby , the goal is to not discard all the learned constraints after the pcnf @xmath117 has been solved .
instead , to the largest extent possible we want to re - use the constraints that were learned from @xmath117 in the process of solving the next pcnf @xmath118 . to this end , the acnf @xmath119 of @xmath118 for @xmath120 , which is maintained by the solver ,
must be initialized with a set @xmath121 of learned clauses and a set @xmath122 of learned cubes such that @xmath123 , @xmath124 and proposition [ prop_learning_satequiv ] holds with respect to @xmath118 .
the sets @xmath125 and @xmath126 contain the clauses and cubes that were learned from the previous pcnf @xmath117 and potentially can be used to derive further constraints from @xmath118 .
if @xmath127 and @xmath128 at the beginning , then the solver solves the pcnf @xmath118 _
incrementally_. for the first pcnf @xmath129 in the sequence , the solver starts with empty sets of learned constraints in the acnf @xmath130 .
each pcnf @xmath118 for @xmath131 in the sequence @xmath132 has the form @xmath133 .
the cnf part @xmath134 of @xmath118 results from @xmath135 of the previous pcnf @xmath136 in the sequence by addition and deletion of clauses .
we write @xmath137 , where @xmath138 and @xmath139 are the sets of deleted and added clauses .
the quantifier prefix @xmath140 of @xmath118 is obtained from @xmath141 of @xmath142 by deletion and addition of variables and quantifiers , depending on the clauses in @xmath139 and @xmath138 .
that is , we assume that the pcnf @xmath118 is closed and that its prefix @xmath140 does not contain superfluous quantifiers and variables . when solving the pcnf @xmath117 using a qcdcl - based qbf solver , learned clauses and cubes accumulate in the corresponding acnf @xmath143 .
assume that the learned constraints are derivable with respect to @xmath117 .
the pcnf @xmath117 is modified to obtain the next pcnf @xmath118 to be solved .
the learned constraints in @xmath144 and @xmath145 might become non - derivable with respect to @xmath118 in the sense of definition [ def_constraint_correct ] .
consequently , proposition [ prop_learning_satequiv ] might no longer hold for the acnf @xmath146 of the new pcnf @xmath118 if previously learned constraints from @xmath144 and @xmath126 appear in @xmath121 and @xmath122 . in this case , the solver might produce a wrong result when solving @xmath118 .
assume that the pcnf @xmath147 has been solved and learned constraints have been collected in the acnf @xmath148 .
the clauses in @xmath138 are deleted from @xmath149 to obtain the cnf part @xmath150 of the next pcnf @xmath133 .
if the derivation of a learned clause @xmath151 depends on deleted clauses in @xmath138 , then we might have that @xmath152 but @xmath153 . in this case , @xmath6 is non - derivable with respect to the next pcnf @xmath118 . hence @xmath6 must be discarded before solving @xmath118 starts so that @xmath154 in the initial acnf @xmath155 . otherwise ,
if @xmath156 then the solver might construct a bogus q - resolution proof for the pcnf @xmath118 and , if @xmath118 is satisfiable , erroneously conclude that @xmath118 is unsatisfiable .
[ ex_clause_delete_incorrect_clause ] consider the pcnf @xmath43 from example [ ex_running ] .
the derivation of the clause @xmath157 shown in fig .
[ fig_derivations ] depends on the clause @xmath158 .
we have that @xmath159 .
let @xmath160 be the pcnf obtained from @xmath43 by deleting @xmath161 .
then @xmath162 because @xmath163 is the only clause which contains the literal @xmath164 .
hence a possible derivation of the clause @xmath157 must use @xmath165 .
however , no such derivation exists in @xmath160 .
there is no clause @xmath64 containing a literal @xmath166 which can be resolved with @xmath165 to produce @xmath167 after a sequence of resolution steps .
consider the pcnf @xmath133 with @xmath168 which is obtained from @xmath169 by _ only adding _ the clauses @xmath139 , but not deleting any clauses . assuming that @xmath170 for all @xmath151 in the acnf @xmath148 , also @xmath171 .
hence all the learned clauses in @xmath144 are derivable with respect to the next pcnf @xmath118 and can be added to the acnf @xmath172 .
like above , let @xmath173 be the acnf of the previously solved pcnf @xmath174 .
dual to clause deletions , the addition of clauses to @xmath149 can make learned cubes in @xmath145 non - derivable with respect to the next pcnf @xmath133 to be solved .
the clauses in @xmath139 are added to @xmath149 to obtain the cnf part @xmath150 of @xmath118 .
an initial cube @xmath175 has been obtained from a model @xmath37 of the previous pcnf @xmath117 , i.e. @xmath176 = \top$ ] .
we might have that @xmath177 \not = \top$ ] with respect to the next pcnf @xmath118 because of an added clause @xmath178 ( and hence also @xmath179 ) such that @xmath180 \not = \top$ ] .
therefore , @xmath37 is not a model of @xmath118 and the initial cube @xmath6 is non - derivable with respect to @xmath118 , i.e. @xmath181 but @xmath182 . hence @xmath6 and every cube whose derivation depends on @xmath6 must be discarded to prevent the solver from generating a bogus cube resolution proof for @xmath118 .
if @xmath118 is unsatisfiable , then the solver might erroneously conclude that @xmath118 is satisfiable .
that is , proposition [ prop_learning_satequiv ] might not hold with respect to non - derivable cubes and the acnf @xmath172 of @xmath118 .
[ ex_clause_add_incorrect_cube ] consider the pcnf @xmath43 from example [ ex_running ] .
the derivation of the cube @xmath95 shown in fig .
[ fig_derivations ] depends on the initial cube @xmath183 , which has been generated from the model @xmath184 .
the cube @xmath76 is derivable with respect to @xmath43 since @xmath185 = \top$ ] , and hence @xmath186 .
the cube @xmath187 is also derivable since @xmath188 .
assume that the clause @xmath189 is added to @xmath43 resulting in the unsatisfiable pcnf @xmath190 .
now @xmath76 is non - derivable with respect to @xmath190 since @xmath191 = \bot$ ] .
further , @xmath192 . consider the pcnf @xmath133 with @xmath193 which is obtained from @xmath169 by _ only deleting _
the clauses @xmath138 , but not adding any clauses .
if after the clause deletions some variable @xmath0 does not occur anymore in the resulting pcnf @xmath118 , then @xmath0 is removed from the quantifier prefix of @xmath118 and from every cube @xmath194 which was learned when solving the previous pcnf @xmath142 .
proposition [ prop_learning_satequiv ] holds for the cleaned up cubes @xmath195 for all @xmath175 with respect to @xmath118 and hence @xmath64 can be added to the acnf @xmath172 .
[ prop_cleaned_up_cubes_sound ] let @xmath196 be the acnf of the pcnf @xmath197 .
let @xmath198 be the pcnf resulting from @xmath142 with @xmath199 , where the variables @xmath200 no longer occur in @xmath134 and are removed from @xmath141 to obtain @xmath140 .
given a cube @xmath201 , let @xmath202 .
proposition [ prop_learning_satequiv ] holds for @xmath64 with respect to @xmath203 : @xmath204 . by induction on the structure of the derivations of cubes in @xmath145 .
let @xmath175 be an initial cube due to the assignment @xmath37 with @xmath176 = \top$ ] . for @xmath205 , we have @xmath206 = \top$ ] since all the clauses containing the variables in @xmath200 were deleted from @xmath117 to obtain @xmath118 .
then the claim holds for the initial cube @xmath207 since @xmath208 .
let @xmath175 be obtained from @xmath209 by existential reduction such that @xmath210 .
assuming that the claim holds for @xmath211 , it also holds for @xmath212 since existential reduction removes existential literals which are maximal with respect to the prefix ordering .
let @xmath175 be obtained from @xmath213 by resolution on variable @xmath0 with @xmath214 , @xmath215 .
assume that the claim holds for @xmath211 and @xmath216 , i.e. @xmath217 and @xmath218 .
if @xmath219 then the claim also holds for @xmath220 with @xmath221 , @xmath222 due to the correctness of resolution ( proposition [ prop_learning_satequiv ] ) . if @xmath223 then the claim also holds for @xmath224 since @xmath225 for all variables @xmath226 , which can be proved by reasoning with tree - like models of qbfs @xcite .
if a variable @xmath0 no longer occurs in the formula , then cubes where @xmath0 has been removed might become non - derivable .
however , due to propositions [ prop_learning_satequiv ] and [ prop_cleaned_up_cubes_sound ] it is sound to keep all the cleaned up cubes ( resolution is not inferentially - complete ) .
moreover , due to the correctness of resolution and existential reduction , proposition [ prop_learning_satequiv ] also holds for new cubes derived from the cleaned up cubes . in practice , the goal is to keep as many learned constraints as possible because they prune the search space and can be used to derive further constraints .
therefore , subsets @xmath123 and @xmath124 of the learned clauses @xmath125 and cubes @xmath126 must be selected so that proposition [ prop_learning_satequiv ] holds with respect to the initial acnf @xmath227 of the pcnf @xmath118 to be solved next .
we describe the implementation of our incremental qcdcl - based solver .
our approach is general and fits any qcdcl - based solver . for incremental
solving we do not apply a sophisticated analysis of variable dependencies by dependency schemes in @xcite . instead , as many other qbf solvers , we use the linear ordering given by the quantifier prefix .
we implemented a stack - based representation of the cnf part of pcnfs based on selector variables and assumptions .
assumptions were also used for incremental qbf - based bmc of partial designs @xcite and are common in incremental sat solving @xcite .
we address the problem of checking which learned constraints can be kept across different solver runs after the current pcnf has been modified . to this end
, we present approaches to check if a constraint learned from the previous pcnf is still derivable from the next one , which makes sure that proposition [ prop_learning_satequiv ] holds .
similar to incremental sat solving , selector variables are used to handle the learned clauses . regarding learned cubes , selector variables can also be used ( although in a way asymmetric to clauses ) , in addition to an alternative approach relying on full derivation dags , which have to be kept in memory . learned cubes might become non - derivable by the deletion of clauses and superfluous variables , but still can be kept due to proposition [ prop_cleaned_up_cubes_sound ] .
we implemented a simple approach which , after clauses have been added to the formula , allows to keep only initial cubes but not cubes obtained by resolution or existential reduction .
let @xmath228 be a pcnf .
we define a set @xmath229 of _ assumptions _ as an assignment such that @xmath230 for all literals @xmath39 .
the variables assigned by @xmath37 are from the first block @xmath231 of @xmath43 . solving the pcnf @xmath43 _ under the set @xmath37 of assumptions
_ amounts to solving the pcnf @xmath44 $ ] .
the definition of assumptions can be applied recursively to the pcnf @xmath44 $ ] .
if @xmath37 assigns all the variables in @xmath231 , then variables from @xmath232 can be assigned as assumptions with respect to @xmath44 $ ] , since @xmath232 is the first block in the quantifier prefix of @xmath44 $ ] .
we implemented the handling of assumptions according to the _ literal - based single instance ( ls ) _ approach ( in the terminology of @xcite ) .
thereby , the assumptions in @xmath37 are treated in a special way so that the variables in @xmath37 are never selected as pivots in the resolution derivation of a learned constraint according to qcdcl - based learning .
similar to sat - solving under assumptions , ls allows to keep all the constraints that were learned from the pcnf @xmath44 $ ] under a set @xmath37 of assumptions when later solving @xmath233 $ ] under a different set @xmath234 of assumptions . in , the cnf part @xmath10 of an acnf @xmath235 to be solved is represented as a stack of clauses .
the clauses on the stack are grouped into _
frames_. the solver api provides functions to push new frames onto the stack , pop present frames from the stack , and to add new clauses to the current topmost frame .
each _ push _
operation opens a new topmost frame @xmath236 .
new clauses are always added to the topmost frame @xmath236 .
each new frame @xmath236 opened by a _ push _
operation is associated with a fresh _ frame selector variable _ @xmath237 .
frame selector variables are existentially quantified and put into a separate , leftmost quantifier block @xmath238 i.e. the current acnf @xmath239 has the form @xmath240 . before a new clause @xmath6
is added to frame @xmath236 , the frame selector variable @xmath237 of @xmath236 is inserted into @xmath6 so that in fact the clause @xmath241 is added to @xmath236 . if all the selector variables are assigned to _
false _ then under that assignment every clause @xmath241 is syntactically equivalent to @xmath6 .
the purpose of the frame selector variables is to _ enable _ or _ disable _ the clauses in the cnf part @xmath149 with respect to the _ push _ and _ pop _ operations applied to the clause stack . if the selector variable @xmath237 of a frame @xmath236 is assigned to _ true
_ then all the clauses of @xmath236 are satisfied under that assignment . in this case , these satisfied clauses are considered disabled because they can not be used to derive new learned clauses in qcdcl .
otherwise , the assignment of _ false _ to @xmath237 does not satisfy any clauses in @xmath236 .
therefore these clauses are considered enabled . before the solving process starts ,
the clauses of frames popped from the stack are disabled and the clauses of frames still on the stack are enabled by assigning the selector variables to _ true _ and _ false _ , respectively .
the selector variables are assigned as assumptions .
this is possible because these variables are in the leftmost quantifier block @xmath238 of the acnf @xmath242 to be solved .
the idea of enabling and disabling clauses by selector variables and assumptions originates from incremental sat solving @xcite .
this approach was also applied to bounded model checking of partial designs by incremental qbf solving @xcite . in
, we implemented the _
push _ and _ pop _ operations related to the clause stack by selector variables similarly to the sat solver @xcite . in the implementation of ,
frame selector variables are maintained entirely by the solver .
depending on the _ push _ and _ pop _ operations , selector variables are automatically inserted into added clauses and assigned as assumptions .
this approach saves the user the burden of inserting selector variables manually into the qbf encoding of a problem and assigning them as assumptions via the solver api .
manual insertion is typically applied in incremental sat solving based on assumptions as pioneered by @xcite .
we argue that the usability of an incremental qbf solver is improved considerably if the selector variables are maintained by the solver .
for example , from the perspective of the user , the qbf encoding contains only variables relevant to the encoded problem . in the following , we consider the problem of maintaining the sets of learned constraints across different solver runs .
as pointed out in section [ sec_inc_satisfiability ] , proposition [ prop_learning_satequiv ] still holds for learned clauses ( cubes ) after the addition ( deletion ) of clauses to ( from ) the pcnf .
therefore , we present the maintenance of learned constraints separately for clause additions and deletions .
a clause @xmath151 in the current acnf @xmath243 might become non - derivable if its derivation depends on clauses in @xmath138 which are deleted to obtain the cnf part @xmath137 of the next pcnf @xmath118 . in , learned clauses in @xmath144 are deleted as follows . as pointed out in the previous section , clauses of popped off frames are disabled by assigning the respective frame selector variables to _
true_. since the formula contains only positive literals of selector variables , these variables can not be chosen as pivots in derivations .
therefore , learned clauses whose derivations depend on disabled clauses of a popped off frame @xmath236 contain the selector variable @xmath237 of @xmath236 .
hence these learned clauses are also disabled by the assignment of @xmath237 .
this approach to handling learned clauses is also applied in incremental sat solving @xcite .
the disabled clauses are physically deleted in a garbage collection phase if their number exceeds a certain threshold .
variables which no longer occur in the cnf part of the current pcnf are removed from the quantifier prefix and , by proposition [ prop_cleaned_up_cubes_sound ] , from learned cubes in @xmath145 to produce cleaned up cubes .
we initialize the set @xmath122 of learned cubes in the acnf @xmath227 of the next pcnf @xmath118 to be solved to contain the cleaned up cubes .
the deletion of learned clauses based on selector variables is not optimal in the sense of definition [ def_constraint_correct ] .
there might be another derivation of a disabled learned clause @xmath6 which does not depend on the deleted clauses @xmath138 .
this observation also applies to the use of selector variables in incremental sat solving .
as illustrated in the context of incremental sat solving , the size of learned clauses might increase considerably due to the additional selector variables @xcite . in the stack - based cnf representation of ,
the clauses associated to a frame @xmath236 all contain the selector variable @xmath237 of @xmath236 .
therefore , the maximum number of selector variables in a new clause learned from the current pcnf @xmath117 is bounded by the number of currently enabled frames .
the sequence of _ push _ operations introduces a linear ordering @xmath244 on the enabled frames @xmath245 and their clauses in the cnf with respect to the point of time where that frames and clauses have been added . in , we implemented the following optimization based on this temporal ordering .
let @xmath6 and @xmath64 be clauses which are resolved in the course of clause learning .
assume that @xmath246 and @xmath247 are the only selector variables of currently enabled frames @xmath245 and @xmath236 in @xmath6 and @xmath64 . instead of computing the usual q - resolvent @xmath248 ,
we compute @xmath249 .
that is , the selector variable of the frame which is smaller in the temporal ordering is discarded from the resolvent . if @xmath250 then the clauses in @xmath245 were pushed onto the clause stack before the clauses in @xmath236 .
the frame @xmath236 will be popped off the stack before @xmath245 .
therefore , in order to properly disable the learned clause @xmath251 after _ pop _ operations , it is sufficient to keep the selector variable @xmath237 of the frame @xmath236 in @xmath251 . with this optimization ,
_ every _ learned clause contains _ exactly one _ selector variable . in the sat solver , an optimization which has similar effects
is implemented .
assume that the pcnf @xmath252 has been solved and that all learned constraints in the acnf @xmath253 of @xmath117 are derivable with respect to @xmath142 .
the set @xmath139 of clauses is added to @xmath149 to obtain the cnf part @xmath137 of the next pcnf @xmath133 . for learned clauses , we can set @xmath254 in the acnf @xmath255 of @xmath118 .
the following example illustrates the effects of adding @xmath139 on the cubes .
[ ex_incorrect_cubes_clean_dag ] consider the cube derivation shown in fig .
[ fig_derivations ] .
as illustrated in example [ ex_clause_add_incorrect_cube ] , the cubes @xmath256 and @xmath95 are non - derivable with respect to the pcnf @xmath190 obtained from @xmath43 by adding the clause @xmath189 .
the initial cube @xmath257 still is derivable because the underlying model @xmath258 of @xmath43 is also a model of @xmath190 .
therefore , when solving @xmath190 we can keep the derivable cubes @xmath77 and @xmath259 .
the non - derivable cubes @xmath76 and @xmath187 must be discarded .
otherwise , qcdcl might produce the cube resolution proof shown in fig .
[ fig_derivations ] when solving the _ unsatisfiable _ pcnf @xmath190 , which is incorrect .
we sketch an approach to identify the cubes in a cube derivation dag @xmath260 which are non - derivable with respect to the next pcnf @xmath261 .
starting at the initial cubes , @xmath260 is traversed in a topological order .
an initial cube @xmath6 is marked as derivable if @xmath262 = \top$ ] , otherwise if @xmath262 \not = \top$ ] then @xmath6 is marked as non - derivable .
this test can be carried out syntactically by checking whether every clause of @xmath118 is satisfied under the assignment given by @xmath6 . a cube @xmath6 obtained by existential reduction or cube resolution
is marked as derivable if all its predecessors in @xmath260 are marked as derivable .
otherwise , @xmath6 is marked as non - derivable .
finally , all cubes in @xmath260 marked as non - derivable are deleted .
the above procedure allows to find a subset @xmath263 of the set @xmath126 of cubes in the solved acnf @xmath235 so that all cubes in @xmath122 are derivable and proposition [ prop_learning_satequiv ] holds for the next acnf @xmath227 .
however , this procedure is not optimal because it might mark a cube @xmath264 as non - derivable with respect to the next pcnf @xmath118 although @xmath265 . [ ex_cube_check_overapprox ] given the satisfiable pcnf @xmath266 , where @xmath267 with the clauses @xmath268 from example [ ex_running ] where @xmath80 , @xmath269 , @xmath82 , @xmath270 , @xmath84 . consider the model @xmath271 of @xmath43 and the initial cube @xmath272 generated from @xmath273 . existential reduction of @xmath274 produces the cube @xmath275 .
assume that the clause @xmath276 is added to @xmath43 to obtain the pcnf @xmath277 .
the initial cube @xmath274 is non - derivable with respect to @xmath277 since @xmath278 \not = \top$ ] .
however , for the cube @xmath279 derived from @xmath274 it holds that @xmath280 .
the assignment @xmath281 is a model of @xmath277 .
let @xmath282 be the initial cube generated from @xmath283 .
then @xmath284 is derivable with respect to @xmath277 . in practice , qcdcl - based solvers typically store only the learned cubes , which might be a small part of the derivation dag @xmath260 , and no edges .
therefore , checking the cubes in a traversal of @xmath260 is not feasible . even if the full dag @xmath260 is available , the checking procedure is not optimal as pointed out in example [ ex_cube_check_overapprox ] .
furthermore , it can not be used to check cubes which have become non - derivable after cleaning up by proposition [ prop_cleaned_up_cubes_sound ] .
hence , it is desirable to have an approach to checking the derivability of _ individual _ learned cubes which is independent from the derivation dag @xmath260 . to this end
, we need a condition which is sufficient to conclude that some _ arbitrary _ cube @xmath6 is derivable with respect to a pcnf @xmath43 , i.e. to check whether @xmath285 . however , we are not aware of such a condition . as an alternative to keeping the full derivation dag in memory ,
a _ fresh _
selector variable can be added to _ each _ newly learned initial cube .
similar to selector variables in clauses , these variables are transferred to all derived cubes .
potentially non - derivable cubes are then disabled by assigning the selector variables accordingly .
however , different from clauses , it must be checked _
explicitly _ which initial cubes are non - derivable by checking the condition in definition [ def_model ] for all initial cubes in the set @xmath145 of learned cubes .
this amounts to an asymmetric treatment of selector variables in clauses and cubes .
clauses are added to and removed from the cnf part by _ push _ and _ pop _ operations provided by the solver api .
this way , it is known precisely which clauses are removed .
in contrast to that , cubes are added to the set of learned cubes @xmath145 on the fly during cube learning . moreover , the optimization based on the temporal ordering of selector variables from the previous section is not applicable to generate shorter cubes since cubes are not associated to stack frames . due to the complications illustrated above , we implemented the following simple approach in to keep only initial cubes . every initial cube computed by the solver
is stored in a linked list @xmath286 of bounded capacity , which is increased dynamically .
the list @xmath286 is separate from the set of learned clauses .
assume that a set @xmath139 of clauses is added to the cnf part @xmath149 of the current pcnf to obtain the cnf part @xmath287 of the next pcnf @xmath261 .
all the cubes in the current set @xmath145 of learned cubes are discarded . for every added clause @xmath288 and for every initial cube @xmath289 , it is checked whether the assignment @xmath37 given by @xmath64 is a model of the next pcnf @xmath118 .
initial cubes @xmath64 for which this check succeeds are added to the set @xmath122 of learned cubes in the acnf @xmath172 of the next pcnf @xmath118 after existential reduction has been applied to them .
if the check fails , then @xmath64 is removed from @xmath286 .
it suffices to check the initial cubes in @xmath286 only with respect to the clauses @xmath288 , and not the full cnf part @xmath134 , since the assignments given by the cubes in @xmath286 are models of the _ current _ pcnf @xmath142 . in the end ,
the set @xmath122 contains only initial cubes all of which are derivable with respect to the acnf @xmath172 . if clauses are removed from the formula , then by proposition [ prop_cleaned_up_cubes_sound ] variables which do not occur anymore in the formula are removed from the initial cubes in @xmath286 . in the incremental qbf - based approach
to bmc for partial designs @xcite , all cubes are kept across different solver calls under the restriction that the quantifier prefix is modified only at the left end .
this restriction does not apply to incremental solving of pcnf where the formula can be modified arbitrarily .
the api of @xcite provides functions to manipulate the prefix and the cnf part of the current pcnf .
clauses are added and removed by the _ push _ and _ pop _ operations described in section [ sec_cnf_stack ] .
new quantifier blocks can be added at any position in the quantifier prefix .
new variables can be added to any quantifier block .
variables which no longer occur in the formula and empty quantifier blocks can be explicitly deleted .
the quantifier block @xmath238 containing the frame selector variables is invisible to the user .
the solver maintains the learned constraints as described in sections [ sec_clause_del ] and [ sec_clause_add ] without any user interaction .
the _ push _ and _ pop _ operations are a feature of . additionally , the api supports the manual insertion of selector variables into the clauses by the user .
similar to incremental sat solving @xcite , clauses can then be enabled and disabled manually by assigning the selector variables as assumptions via the api . in this case , these variables are part of the qbf encoding and the optimization based on the frame ordering presented in section [ sec_clause_del ] is not applicable . after a pcnf has been found unsatisfiable ( satisfiable ) under assumptions where the leftmost quantifier block is existential ( universal ) , the set of relevant assumptions which were used by the solver to determine the result can be extracted .
& _ discard lc _ & _ keep lc _ & _
diff.(% ) _ + @xmath290 : & @xmath291 & @xmath292 & -10.88 + @xmath293 : & 3,833,077 & 2,819,492 & -26.44 + @xmath294 : & 139,036 & 116,792 & -16.00 + @xmath295 : & 8,243 & 6,360 & -22.84 + @xmath296 : & 99.03 & 90.90 & -8.19 + @xmath297 : & 28.56 & 15.74 & -44.88 + & _ discard lc _ & _ keep lc _ & _
diff.(% ) _ + @xmath290 : & @xmath298 & @xmath299 & -14.40 + @xmath293 : & @xmath300 & @xmath301 & -3.62 + @xmath294 : & 117,019 & 91,737 & -21.61 + @xmath295 : & 10,322 & 8,959 & -13.19 + @xmath296 : & 100.15 & 95.36 & -4.64 + @xmath297 : & 4.18 & 2.83 & -32.29 + & _ discard lc _ & _ keep lc _ & _
diff.(% ) _ + @xmath290 : & @xmath302 & @xmath303 & -86.62 + @xmath293 : & 186,237 & 15,031 & -91.92 + @xmath294 : & 36,826 & 1,228 & -96.67 + @xmath295 : & 424 & 0 & -100.00 + @xmath296 : & 21.94 & 4.32 & -79.43 + @xmath297 : & 0.75 & 0.43 & -42.66 + & _ discard lc _ &
_ keep lc _ & _
diff.(% ) _ + @xmath290 : & @xmath304 & @xmath305 & -77.94 + @xmath293 : & 103,330 & 8,199 &
-92.06 + @xmath294 : & 31,489 & 3,350 & -89.37 + @xmath295 : & 827 & 5 & -99.39 + @xmath296 : & 30.29 & 9.78 & -67.40 + @xmath297 : & 0.50 & 0.12 & -76.00 + to demonstrate the basic feasibility of general incremental qbf solving , we evaluated our incremental qbf solver based on the instances from _
qbfeval12 second round ( sr ) _ with and without preprocessing by .
we disabled the sophisticated dependency analysis in terms of dependency schemes in and instead applied the linear ordering of the quantifier prefix in the given pcnfs . for experiments
, we constructed a sequence of related pcnfs for _ each _ pcnf in the benchmark sets as follows .
given a pcnf @xmath43 , we divided the number of clauses in @xmath43 by 10 to obtain the size of a slice of clauses .
the first pcnf @xmath129 in the sequence contains the clauses of one slice .
the clauses of that slice are removed from @xmath43 .
the next pcnf @xmath160 is obtained from @xmath129 by adding another slice of clauses , which is removed from @xmath43 .
the other pcnfs in the sequence @xmath306 are constructed similarly so that finally the last pcnf @xmath307 contains all the clauses from the original pcnf @xmath43 . in our tests , we constructed each pcnf @xmath117 from the previous one @xmath308 in the sequence by adding a slice of clauses to a new frame after a _ push _ operation .
we ran on the sequences of pcnfs constructed this way with a wall clock time limit of 1800 seconds and a memory limit of 7 gb .
tables [ tab_experiments_eval2012 ] and [ tab_experiments_eval2012_pop ] show experimental results on sequences @xmath309 of pcnfs and on the reversed ones @xmath310 , respectively .
to generate @xmath311 , we first solved the sequence @xmath312 and then started to discard clauses by popping the frames from the clause stack of depqbf via its api . in one run ( _
discard lc _ ) , we always discarded all the constraints that were learned from the previous pcnf @xmath117 so that the solver solves the next pcnf @xmath118 ( @xmath308 with respect to table [ tab_experiments_eval2012_pop ] ) starting with empty sets of learned clauses and cubes . in another run ( _
keep lc _ ) , we kept learned constraints as described in sections [ sec_clause_del ] and [ sec_clause_add ] . this way ,
70 out of 345 total pcnf sequences were fully solved from the set _
qbfeval12-sr _ by both runs , and 112 out of 276 total sequences were fully solved from the set _ qbfeval12-sr - bloqqer_. the numbers of assignments , backtracks , and wall clock time indicate that keeping the learned constraints is beneficial in incremental qbf solving despite the additional effort of checking the collected initial cubes . in the experiment reported in table [ tab_experiments_eval2012 ] clauses
are always added but never deleted to obtain the next pcnf in the sequence .
thereby , across all incremental calls of the solver in the set _
qbfeval12-sr _ on average 224 out of 364 ( 61% ) collected initial cubes were identified as derivable and added as learned cubes . for
the set _ qbfeval12-sr - bloqqer _ , 232 out of 1325 ( 17% ) were added .
related to table [ tab_experiments_eval2012_pop ] , clauses are always removed but never added to obtain the next pcnf to be solved , which allows to keep learned cubes based on proposition [ prop_cleaned_up_cubes_sound ] . across all incremental calls of the solver in the set _
qbfeval12-sr _ on average 820 out of 1485 ( 55% ) learned clauses were disabled and hence effectively discarded because their q - resolution derivation depended on removed clauses .
for the set _ qbfeval12-sr - bloqqer _ , 704 out of 1399 ( 50% ) were disabled .
we presented a general approach to incremental qbf solving which integrates ideas from incremental sat solving and which can be implemented in any qcdcl - based qbf solver .
the api of our incremental qbf solver provides _ push _ and _ pop _ operations to add and remove clauses in a pcnf .
this increases the usability of our implementation .
our approach is application - independent and applicable to arbitrary qbf encodings .
we illustrated the problem of keeping the learned constraints across different calls of the solver . to improve cube learning in incremental qbf solving ,
it might be beneficial to maintain ( parts of ) the cube derivation in memory .
this would allow to check the cubes more precisely than with the simple approach we implemented .
moreover , the generation of proofs and certificates @xcite is supported if the derivations are kept in memory rather than in a trace file .
dual reasoning @xcite and the combination of preprocessing and certificate extraction @xcite are crucial for the performance and applicability of cnf - based qbf solving .
the combination of incremental solving with these techniques has the potential to further advance the state of qbf solving .
our experimental analysis demonstrates the feasibility of incremental qbf solving in a general setting and motivates further applications , along with the study of bmc of partial designs using incremental qbf solving @xcite .
related experiments with conformant planning based on incremental solving by showed promising results @xcite .
further experiments with problems which are inherently incremental can provide more insights and open new research directions .
audemard , g. , lagniez , j.m .
, simon , l. : improving glucose for incremental sat solving with assumptions : application to mus extraction . in : jrvisalo , m. , van gelder , a. ( eds . ) sat .
lncs , vol .
7962 , pp .
309317 . springer ( 2013 ) cashmore , m. , fox , m. , giunchiglia , e. : planning as quantified boolean formula . in : raedt , l.d . ,
bessire , c. , dubois , d. , doherty , p. , frasconi , p. , heintz , f. , lucas , p.j.f .
frontiers in artificial intelligence and applications , vol .
242 , pp . 217222 .
ios press ( 2012 ) hillebrecht , s. , kochte , m.a . ,
erb , d. , wunderlich , h.j . , becker , b. : accurate qbf - based test pattern generation in presence of unknown values . in : macii ,
e. ( ed . ) date .
. 436441 . eda consortium san jose , ca , usa / acm dl ( 2013 ) lonsing , f. , egly , u. : incremental qbf solving by depqbf ( extended abstract ) . in : the 4th international congress on mathematical software , icms 2014 , seoul ,
korea , august 2014 . proceedings .
lncs , vol .
springer ( 2014 ) , to appear .
lonsing , f. , egly , u. , van gelder , a. : efficient clause learning for quantified boolean formulas via qbf pseudo unit propagation . in : jrvisalo ,
m. , van gelder , a. ( eds . ) sat .
lncs , vol . 7962 , pp .
100115 . springer ( 2013 ) marin , p. , miller , c. , becker , b. : incremental qbf preprocessing for partial design verification - ( poster presentation ) . in : cimatti , a. , sebastiani , r. ( eds . ) sat .
lncs , vol . 7317 , pp .
473474 . springer ( 2012 ) niemetz , a. , preiner , m. , lonsing , f. , seidl , m. , biere , a. : resolution - based certificate extraction for qbf - ( tool presentation ) . in : cimatti ,
a. , sebastiani , r. ( eds . ) sat .
lncs , vol . 7317 , pp .
430435 . springer ( 2012 ) silva , j.p.m .
, lynce , i. , malik , s. : conflict - driven clause learning sat solvers . in : biere ,
a. , heule , m. , van maaren , h. , walsh , t. ( eds . ) handbook of satisfiability , faia , vol .
ios press ( 2009 ) zhang , l. , malik , s. : towards a symmetric treatment of satisfaction and conflicts in quantified boolean formula evaluation . in : hentenryck , p.v .
( ed . ) cp .
lncs , vol . 2470 , pp .
springer ( 2002 ) | we consider the problem of incrementally solving a sequence of quantified boolean formulae ( qbf ) .
incremental solving aims at using information learned from one formula in the process of solving the next formulae in the sequence . based on a general overview of the problem and related challenges , we present an approach to incremental qbf solving which is application - independent and hence applicable to qbf encodings of arbitrary problems .
we implemented this approach in our incremental search - based qbf solver and report on implementation details .
experimental results illustrate the potential benefits of incremental solving in qbf - based workflows . |
studies aimed at understanding the consequences of the interaction of electromagnetic fields with metal , semiconductor and molecular nanostructures under the effort to construct , characterize , manipulate and control plasmonic devices @xcite . recurring themes in these studies are the plasmonic response of aggregates of nano - particles @xcite , and the possibility to transmit electromagnetic energy over constrictions substantially smaller than the radiation wavelength .
composite structures ( metal - dielectric , metal - semiconductor ) are often found useful because light can localize at their interfaces .
of particular importance are metal- molecule composites where strong plasmon - exciton coupling together with the non - linear optical response of the molecular system can generate new physical behavior on one hand and provide more control capabilities by tuning the molecular subsystem on the other . in this paper
we study a class of systems of the latter kind , focusing on the light transmission properties of a model system comprising metal rods and molecular aggregates of sub - wavelength dimensions .
transmission properties of nano - size structure are usually discussed in two connotations .
first , following early studies by gersten and nitzan @xcite there is substantial interest in the way plasmonic particles affect excitation energy transfer in molecular systems @xcite .
second , plasmonic structure can operate as sub - wavelength waveguides which are important in the construction of light controlled nano - devices and nano - size optical communication and information storage systems @xcite . for the latter systems two principal design systems
have been studied : one comprises a chain of nanoparticles @xcite , where waveguiding is achieved by plasmon - hopping between nanoparticles and energy transport may be regarded as motion along a plasmonic band .
the other uses the optical properties of metal- dielectric ( including metal - vacuum ) interfaces @xcite .
molecular aggregates were used in both design types . in refs .
@xcite they are used as reporters for the distribution of electromagnetic energy along the waveguide , while in refs .
@xcite they constitute an active constituent of the dielectric subsystem . another way in which a molecular component can affect the operation in devices based on plasmon hybridization and hopping is in affecting the hybridization characteristics of the system .
obviously , the interactions between plasmonic excitations on different particles will depend on the molecular environment between them @xcite .
when the molecular optical response is far from resonance with the relevant plasmonic frequencies this dependence may be accounted for by incorporating a host medium with a suitable dielectric constant . in the linear response regime
a frequency dependent dielectric function can represent a molecular system in resonance with the plasmonic spectrum , however such a procedure may not properly account for the distinction between lifetime and dephasing relaxation processes .
alternatively , the molecular subsystem is described here explicitly and quantum mechanically using a variant of the procedure described in ref .
@xcite ( see also refs .
@xcite ) . this makes it possible to address the regime of _ strong exciton - plasmon coupling _
, where the molecular species does not only modify the plasmon - plasmon coupling but becomes an active component of the transmitting system .
we note in passing that the latter procedure can also describe the molecular system in the non - linear response regime @xcite , although we do not address this regime in the present study .
the effect of strong exciton - plasmon coupling on the optical response properties of metal - molecules composites have been under active study for some time @xcite .
here we focus on its manifestation in the electromagnetic energy- transmission properties of a system comprising two metal cylinders aligned along a common axis with a gap of variable length between them .
light is injected into the system using a source point dipole located at one end of the two metal - rod system , and the transmitted intensity is recorded at the other end .
the effect of a molecular aggregate filling the gap between the metal cylinders is studied in order to elucidate the role of several key parameters that characterize the molecular species ( a ) molecular density , ( b ) molecular transition frequency and ( c ) molecular lifetime and dephasing relaxation rates .
we note that some issues related to this study were addressed in previous works . on the experimental side , benner et al @xcite
have studied plasmon transmission ( and electronic conduction ) across small constrictions between metal wires , highlighting the need to account for heating and thermal expansion effects in realistic experiments .
neuhauser and coworkers @xcite have placed a single molecule between two metal spheres in order to simulate its effect on the plasmon - plasmon coupling . while their single molecule model uses a similar density matrix description that can study the role of molecular relaxation processes , these issues where not explicitly addressed in these studies .
our earlier work @xcite and a recent work by sadeghi @xcite have addressed the role of molecular dephasing in the coherent response of strongly coupled exciton - plasmon systems .
gu and coworkers @xcite have recently considered the transmission of electromagnetic signal across a gap in a configuration similar to the one studied here , however without molecules .
the present study is aimed to elucidate the way such transmitted energy is affected by strong exciton - plasmon coupling .
our system consists of two metal cylinders of lengths @xmath1 and @xmath2 and equal diameters @xmath3 , lying along a common axis and separated by a gap of width @xmath4 ( see the inset of fig .
( [ fig1 ] ) ) .
this gap can be bridged by a molecular aggregate of the same diameter - an assembly of two - level point objects whose optical response is described by the optical bloch equations @xcite . for comparison , we also consider the corresponding system with a vacuum gap , and the system with no gap ( @xmath4 ) .
the dynamics of the electric , @xmath5 , and magnetic , @xmath6 , fields is simulated using classical maxwell s equations [ maxwell ] @xmath7 where @xmath8 and @xmath9 are the permittivity and the permeability of free space , respectively . the current source in the ampere law ( [ maxwell]b ) , @xmath10 ,
corresponds to either the current density in spatial regions occupied by metal ( eq .
( [ drude_j ] ) below ) or the macroscopic polarization current , @xmath11 , in space filled with a molecular aggregate .
the dispersion of metal is taken into account via drude model with the dielectric constant of metal , @xmath12 , in the form @xmath13 where @xmath14 is the damping parameter , @xmath15 is the bulk plasma frequency , and @xmath16 is the high - frequency limit of the dielectric constant . for the range of frequencies considered in this work the following set of parameters was chosen to represent gold : @xmath17 , @xmath18 ev , and @xmath19 ev @xcite . the corresponding current density in the metal region is evaluated according to the following equation @xcite @xmath20 the optical response of the molecular subsystem is simulated using rate equations for a two - level system driven by a local electric field @xmath5 @xcite [ rate_equations ] @xmath21 where @xmath22 and @xmath23 ( @xmath24 , where @xmath25 is the number density of molecules ) correspond to the populations of the ground and the excited molecular states , respectively , @xmath26 is the macroscopic polarization , @xmath27 is the radiationless decay rate of the excited state , @xmath28 is the pure dephasing rate , and @xmath29 is the energy separation of the molecular levels .
the coupling constant @xmath30 is given by [ ] @xmath31 where @xmath32 is the transition dipole moment .
( [ rate_equations ] ) and ( [ constants ] ) assume that the molecular optical response is isotropic . consequently , the orientation of the local induced polarization is along the polarization of the local electric field . in the standard situation where the individual molecule does not respond isotropically , the results reported below correspond to the assumption that the distribution of molecular orientations is isotropic
this aspect of the model can be generalized in order to investigate the interesting possibility that molecular effects in plasmon transport can be affected by the molecular orientation . in this work
we do not consider such effects .
the system of coupled equations ( [ maxwell ] ) , ( [ drude_j ] ) , and ( [ rate_equations ] ) is solved numerically on a multi - processor computer @xcite .
the space is discretized in accordance with fdtd algorithm @xcite in three dimensions .
the spatial resolution of @xmath33 nm is chosen to achieve numerical convergence and avoid staircase effects ( artificial `` _ _ hot _ _ '' spots in the local field due to discretization of curved surfaces in cartesian coordinates ) .
the time step is @xmath34 as , where @xmath35 is the speed of light in vacuum .
open boundaries are simulated using convolutional perfectly matched layers ( cpml ) @xcite .
we found that for a system considered here the best results were achieved with @xmath36 cpml layers .
the total simulation domain for all calculations is @xmath37 .
we employ short - pulse method ( spm ) @xcite to calculate linear response of the plasmonic / excitonic system .
the time envelope of the probe incident pulse , @xmath38 , is taken in the form of the blackman - harris window @xmath39 the pulse duration is denoted as @xmath40 , other parameters are : @xmath41 , @xmath42 , @xmath43 , and @xmath44 .
the results discussed in the next section are obtained for @xmath45 fs pulse duration leading to an incident pulse with nearly flat spectrum for frequencies between @xmath46 ev and @xmath47 ev . unless otherwise stated , for the calculations reported below we employ two cylindrical identical metal wires with dimensions @xmath48 nm , and @xmath49 nm with a small empty gap in between .
the system is excited locally by a pointwise soft source ( classical point dipole ) of the functional form given by eq .
( [ blackman - harris ] ) that is placed @xmath50 nm from the rightmost wire .
simulations with the driving dipole directed along the @xmath51-axis results in significantly higher transmission compared to any transverse polarization and we have therefore used this driving polarization .
the transmitted intensity in the @xmath51 direction , @xmath52 , is detected on the opposite side also at a distance of @xmath50 nm from the leftmost wire at a given point on the grid .
it should be noted that the utilization of spm implies that we consider only elastic scattering . for a case of interacting wires without molecules
it is obviously the case as the optical response of metal is treated linearly using drude model . in case of molecules one has to keep the incident peak amplitude low enough such that the population of the molecular excited state is always significantly smaller than @xmath46 . in the simulations discussed below this condition
was carefully monitored at all times .
the structure composed of two closely spaced wires shown in the inset of fig . [ fig1 ] is examined first with an empty gap between the wires .
distinct surface plasmon - polariton ( spp ) modes : longitudinal mode associated with oscillations of a charge density along the @xmath51-axis and transverse mode corresponding to oscillations in the direction perpendicular to @xmath51-axis @xcite .
the inset schematically depicts two closely spaced wires with lengths @xmath53 separated by a subwavelength gap @xmath4 .
the excitation and detection are carried out locally at a distance of @xmath50 nm from the wires surface with the source dipole oscillating in the @xmath51 direction .
all simulations in this paper are carried out for wires with a circular cross - section of @xmath54 nm in diameter .
the main panel shows the transmitted intensity in the longitudinal direction , @xmath55 , as a function of the incident frequency .
the black line shows data for a single wire with @xmath56 nm .
other lines are calculated for two closely space identical wires with @xmath48 nm and an empty gap of : @xmath57 nm ( red short dashed line ) , @xmath58 nm ( green long dashed line ) , @xmath54 nm ( blue dash - dotted line ) , and @xmath59 nm ( magenta dash - dot - dotted line).,scaledwidth=90.0% ] consider first the transmission characteristics of a single wire ( @xmath56 nm ) . in the spectral regime
displayed three peaks are shown . by examining their wire - length dependence and the corresponding charge densities @xcite the lower frequency peak at @xmath60 ev is identified as a longitudinal dipolar plasmon , the next one at @xmath61 ev is found to be a longitudinal quadrupole plasmon while the peak at @xmath62 ev corresponds to the transverse dipolar plasmon . for a single wire of length 100 nm the longitudinal dipole mode peaks at @xmath63 ev
, the longitudinal quadrupolar mode has moved to higher frequency above the range shown , while the transverse mode remains at @xmath62 ev . with a small gap separating the two wires one can expect to see a manifestation of the interaction between two longitudinal spp modes supported by each wire .
as two identical wires support modes with the same frequency , the close proximity of such modes permits the energy exchange between wires and thus lifts the degeneracy .
this effect is clearly seen in fig .
[ fig1 ] as a splitting of the longitudinal spp mode at @xmath63 ev .
the splitting is noticeable for a gap of @xmath57 nm since spp modes are evanescent and decay exponentially with a distance from the surface of each wire , the gap becomes narrower the splitting significantly increases reaching @xmath64 mev for a @xmath59 nm gap .
the observed rabi splitting indicates the fact that interacting longitudinal spp modes are in the so - called strong coupling regime permitting efficient energy exchange between wires @xcite . by comparing transmission through a single wire ( black solid line ) and through two closely spaced wires with a gap of @xmath59 nm ( magenta dash - dot - dotted line )
we observe nearly perfect overlap of the quadrupole mode for a single wire with the high frequency mode in the system of coupled wires .
similar simulations comparing transmission through wider gaps with that through a single wire with a corresponding length confirm the quadrupole nature of the high frequency mode .
it should be noted that the transmission in vacuum if wires are removed is several orders of magnitude smaller than that shown in fig.[fig1 ] .
panel ( a ) shows the transmitted intensity in the longitudinal direction , @xmath55 , as a function of the incident frequency , for the system of two wires with the fixed @xmath65 nm and variable @xmath1 and a gap of @xmath66 nm between the wires ( see the inset of fig .
[ fig1 ] for details of the geometry ) .
the black solid line shows results for symmetric system with @xmath67 nm , red dashed line shows data for @xmath68 nm , green dash - dotted line shows results for @xmath69 nm , and blue dash - dot - dotted line shows results for @xmath70 nm . panel ( b ) shows the energies of two transmission resonances as functions of the wire s
length @xmath1 .
horizontal black dashed line indicates the energy of the longitudinal spp resonance for a single @xmath71 nm wire .
green dash - dotted line shows how the energy of the longitudinal spp resonance of a single wire varies with its length .
other parameters are the same as in fig .
[ fig1].,scaledwidth=90.0% ] to further scrutinize the strong coupling between interacting longitudinal spp modes we perform a series of simulations varying the length of one wire while keeping the length of the other constant .
this allows us to alter the frequency of one of the longitudinal spp resonances sweeping through the other .
the results of the simulations are shown in fig .
we note that the high energy spp mode seen in fig .
[ fig2]a at @xmath72 ev is the transverse spp resonance position changes mildly from @xmath73 ev to @xmath74 ev when the length of one of the wires varies .
the structure of the splitting of the longitudinal mode spectrum changes significantly , with the higher frequency peak changing much faster than the lower one .
this apparent asymmetry is easily explained by referring to the avoided crossing behavior shown in fig .
[ fig2]b : when the length of the wire is comparable to its diameter it scatters light as a nanoparticle with a single dipolar mie resonance of frequency similar to the transverse mode ( @xmath72 ev ) . as the length increases this mode moves to the red while the unperturbed mode of the other wire of fixed length is unchanged .
an avoided crossing is seen when the two lengths match .
steady - state electromagnetic intensity distributions for the system of two interacting wires separated by an empty gap of @xmath59 nm ( see the inset in fig .
[ fig1 ] for details of the geometry ) .
the length of each wire is @xmath71 nm . the system is excited by a pointwise dipole placed at the right side and oscillating in longitudinal @xmath51-direction .
panel ( a ) shows normalized electromagnetic intensity as a function of @xmath75 and @xmath51 ( in nm ) for the incident frequency of @xmath76 ev ( the low energy resonance for data shown as a magenta dash - dot - dotted line in fig .
( [ fig1 ] ) ) .
panel ( b ) shows normalized electromagnetic intensity calculated @xmath77 ev ( the high energy resonance for data shown as a magenta dash - dot - dotted line in fig .
[ fig1 ] ) .
the electromagnetic intensity distributions are plotted in logarithmic scale and normalized to the incident intensity.,scaledwidth=90.0% ] several other observations can be made .
first , although the transversal modes of the two wires are in principle also coupled , this coupling is too weak to give a noticeable rabi splitting .
second , even at the shortest length similar to the diameter , the plasmon frequency associated with the short wire is affected by the presence of the other wire and slightly deviate from the value obtained for the isolated wire ( green dash - dotted line in fig .
[ fig2]b ) .
this probably results from the damping and frequency shift affected by the proximity to a dissipative dielectric object ( the second wire ) but may also reflect the effect of interaction with the transversal plasmon of this second wire .
as expected , similar results were obtained when we varied @xmath2 with @xmath1 being kept constant .
finally we note that these observations may be sensitive to the polarization of the injecting source .
we defer the study of this issue to later work . in order to understand the physics behind the rabi splitting
we examine spatial distributions of the electromagnetic energy at two resonant frequencies . in the strong coupling regime
the system of two wires forms two states as illustrated in fig .
the low energy mode has a maximum in the gap while the high energy mode has a node .
obviously we are seeing the equivalent of the behavior of bonding and antibonding orbitals in a system of two coupled emitters @xcite as this was confirmed in fig .
[ fig1 ] by comparing transmission through a single wire with that through two closely spaced wires .
coupled plasmon - exciton system .
panel ( a ) shows transmission as a function of the incident frequency calculated for two wires with a length of @xmath48 nm with a molecular aggregate placed in the gap between the wires ( @xmath66 wide ) .
the molecular transition frequency @xmath78 ev corresponds to the solid and dotted black lines ( see below ) , dashed red line shows data for @xmath79 ev , dash - dotted green line shows results for @xmath80 ev , and dash - dot - dotted blue line shows data for @xmath81 ev .
the molecular number density is @xmath82 m@xmath83 for all lines except for the dotted black line , which is calculated at @xmath84 m@xmath83 .
panel ( b ) shows transmission as a function of the incident frequency and molecular transition frequency , @xmath85 .
two horizontal red lines show the energies of the hybrid sp modes without molecules in the gap .
the third red line with a slope shows the molecular transition frequency sweep.,scaledwidth=90.0% ] next we add a molecular aggregate filling the gap between the interacting wires and investigate how molecules resonant to longitudinal spp modes modify the transmission spectrum .
the values of the molecular parameters used in this work are : the molecular transition dipole @xmath86 debye , @xmath87 ev ( corresponding to @xmath88 ps ) , @xmath89 ev ( corresponding to @xmath90 fs ) . when a molecular aggregate is placed inside the gap the local electromagnetic field
is coupled to molecules .
this system is analogous to three coupled oscillators : two longitudinal spp electromagnetic modes and molecular excitons .
the ratio of intensities spatially averaged over gap s volume for the low energy mode to the high one for the parameters in fig .
[ fig3 ] is @xmath91 .
the coupling strength depends on a local field amplitude .
one can anticipate that the low energy plasmon mode with a higher amplitude in the gap would couple to molecules appreciably stronger than the mode at the higher energy . as noted above
, the antibonding mode is close in frequency to the quadrupolar mode of a single wire of similar total length .
[ fig4 ] shows results of simulations with varying molecular transition energy , @xmath85 . with molecules having a transition close to the low energy mode
the transmission exhibits three resonances with actual transmission dropping by more than @xmath47 orders of magnitude for the incident photon energy slightly higher than the molecular transition ( black solid line , fig .
[ fig4]a , @xmath92 ev ) .
additional simulations ( not shown ) were carried out to calculate a spatial distribution of the electromagnetic energy for parameters corresponding to very low transmission .
we found that in this case the molecular aggregate acts as a very efficient absorber dropping the overall transmission .
when the molecular transition energy is around @xmath63 ev the transmission exhibits signs of interference effects as its shape has a clear fano lineshape ( green dash - dotted line , fig .
[ fig4]a ) .
however at energies close to the high energy mode near @xmath93 ev the transmission looks nearly unperturbed compared to the case without molecules .
this can be explained if we recall the fact that the high energy mode has a node in the gap making the coupling with molecules negligibly small .
[ fig4]b shows results of the fine sweep over the molecular transition frequency , @xmath85 .
since the molecular aggregate is placed in the gap between two wires one would expect to observe coupling between the low frequency sp mode ( mainly localized in the gap ) and nearly no effect on the antibonding mode .
such a regime is manifested by a clear avoided crossing near @xmath94 ev with the rabi splitting reaching about @xmath95 mev .
the effect of resonant molecules on the antibonding mode located near @xmath77 ev is not trivial .
although there is no strong coupling observed molecules act as absorbers resulting in a narrow minimum occurring inside the transmission maximum ( see blue dash - dot - dotted line in fig . [
fig4]a ) .
the @xmath51-component of the poynting vector multiplied by the distance from the axis connecting wires , @xmath96 , as a function of @xmath97 .
@xmath98 is calculated at the center of the system at @xmath99 .
black solid line shows results for a single wire @xmath100 nm long calculated at @xmath101 ev ( see black line in fig .
[ fig1 ] ) .
red dashed line presents results for two wires with @xmath48 nm and an empty gap of @xmath66 nm calculated at @xmath102 ev ( see magenta dash - dot - dotted line in fig .
[ fig1 ] ) .
green dash - dot - dotted line shows results for two wires with the same characteristics as for the red line but with a gap filled with a non - dispersive dielectric with @xmath103 calculated at @xmath104 ev ( corresponding to the maximum transmission ) .
blue dash - dotted line shows results for the same system of two wires with the gap filled with molecules .
the latter is calculated at @xmath105 ev ( see black solid line in fig . [ fig4]a ) .
magenta line connecting circles shows results for two wires with @xmath48 nm and gap of @xmath66 nm submerged in molecular cylinder of the radius @xmath54 nm .
it is calculated at @xmath106 ev .
molecules for all calculations are resonant at @xmath107 ev and the number density is @xmath108 m@xmath83 .
all wires are @xmath54 nm in diameter.,scaledwidth=90.0% ] to investigate how electromagnetic energy is transported through a system of closely spaced nanowires we performed series of simulations varying the physical environment of the gap .
first set of simulations was carried out with non - resonant molecules filling the gap .
we found that there is almost no effect on transmission if the molecular transition energy is below @xmath109 ev irrespective of how high the molecular concentration is .
this suggested another test , namely to replace molecules with a perfect electric conductor ( pec ) completely closing the gap between the wires .
simulations showed that the effect of pec on transmission is a red - shift of the low energy mode while the energy of the `` antibonding '' state remains constant .
for the system with @xmath66 nm and @xmath48 nm the shift of the low energy mode is @xmath110 ev .
these findings indicate that the electromagnetic energy transport occurs primarily along the surface of wires .
additional simulations examining spatial distributions of the poynting vector confirmed that hypothesis as illustrated in fig .
the @xmath51-component of the poynting vector is evaluated for several representative cases as a function of the radial distance from the symmetry axis connecting two wires .
we note that the negative sign of the energy flux corresponds to the propagation of em energy from the source towards the detector ( see the inset in fig .
[ fig1 ] ) .
the magenta line shows simulations performed for a system comprised of two wires fully submerged in a cylinder composed of molecules .
this system exhibit several resonances in the transmission spectrum ( not shown ) .
the `` bonding '' mode ( see fig .
[ fig1 ] , blue dash - dotted line , resonance at @xmath111 ev , see also fig .
[ fig3]b ) splits into two modes due to strong coupling between molecules and the plasmon mode .
the frequency at which the energy flux is evaluated corresponds to the maximum transmission at @xmath112 ev .
one can clearly see the em energy mostly localized in the molecular layer @xmath59 nm @xmath113 nm . all five cases clearly demonstrate the fact that the propagation mainly occurs along the surface of wires .
finally , the following observation from our simulation may appear surprising : the result shown in fig .
[ fig4 ] are not sensitive to the nature of the relaxation processes modeled in eqs .
( [ rate_equations ] ) in the following sense : if the magnitudes of the population and dephasing relaxation rates are varied such that @xmath114 the same spectrum is produced .
this behavior characterizes systems in which spontaneous emission does not play a significant part in the spectral response ( in the present calculation using classical em field it is simply ignored ) . in such systems the optical response is dominated by the classical molecular polarization which is sensitive only to the sum @xmath115 .
using a numerical scheme based on the coupled maxwell - bloch equations implemented within the fdtd electromagnetic numerical solver , we have studied the transfer of electromagnetic radiation across a molecular gap separating two co - axial metal cylinders . in the absence of the molecular spacer in the gap separating the rods this system exhibits the consequence of plasmon - plasmon coupling , in particular between longitudinal plasmons in the two rods .
the nature and magnitude of this coupling can be studied by varying the gap width as well as by changing the length of one wire keeping the other fixed , which produces the familiar non - crossing behavior . as a function of frequency
the transmission is dominated by a splitted longitudinal plasmon peak .
the two hybrid modes are the dipole - like `` bonding '' mode characterized by a peak intensity in the gap , and a quadrupole - like `` antibonding '' mode whose amplitude vanishes at the gap center . when off resonant @xmath0level molecules are placed in the gap , almost no effect on the frequency dependent transmission is observed .
we have traced this behavior to the observation that much of the transmission takes place along the cylinders edge .
the `` bonding '' mode is significantly affected by the dielectric properties of the gap .
in contrast , when the molecular system is in resonance with the plasmonic lineshape , the transmission is strongly modified , showing characteristics of strong exciton - plasmon coupling , modifying mostly the transmission near the lower frequency `` bonding '' plasmon mode .
it is interesting to note that the presence resonant molecules species in the gap affects not only the molecule - field interaction but also the spatial distribution of the field intensity and the electromagnetic energy flux across the junction .
this study can be extended in several ways .
first and obvious is the fact that the observation reported here can be sensitive to the molecular orientation in the gap . in the present study
we have assumed that this orientation is random , implying an average isotropic molecular response , and a preferred molecular orientation may be an important parameter in determining the transmission across the junction . a more subtle issue is the role played by molecular fluorescence . in present experimental studies
the fluorescence signal from dye molecules placed along the nano waveguide is used to report on the electromagnetic field distribution along the guide @xcite . in the configuration used in the present study spontaneous emission by strongly fluorescent molecule may play an active role in the observed optical response . to study
such one needs to go beyond the current level of description that treat the electromagnetic field completely classically .
we defer such considerations to future work .
both authors acknowledge support by the collaborative bsf grant no . 2014113 .
m.s . is also grateful for financial support by afosr grant no .
fa9550 - 15 - 1 - 0189 .
the research of a.n .
is also supported by the israel science foundation .
93ifxundefined [ 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 _ _ , vol .
( , ) link:\doibase 10.1021/jp026770 + [ * * , ( ) ] http://stacks.iop.org/0953-8984/17/i=12/a=003 [ * * , ( ) ] link:\doibase 10.1021/nl050928v [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/jp063879z [ * * , ( ) ] link:\doibase 10.1021/nl071008a [ * * , ( ) ] http://stacks.iop.org/0953-8984/19/i=9/a=096210 [ * * , ( ) ] link:\doibase 10.1021/jp7120356 [ * * , ( ) ] link:\doibase 10.1021/nl803621x [ * * , ( ) ] link:\doibase 10.1021/jp810646 m [ * * , ( ) ] link:\doibase 10.1021/jp911371r [ * * , ( ) ] link:\doibase 10.1021/ja104174 m [ * * , ( ) ] link:\doibase 10.1021/jp100861w [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.83.075426 [ * * , ( ) ] link:\doibase 10.1021/jp205158w [ * * , ( ) ] link:\doibase 10.1021/jp205055h [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.84.165447 [ * * , ( ) ] link:\doibase 10.1021/jp302013k [ * * , ( ) ] link:\doibase 10.1021/ar200337u [ * * , ( ) ] link:\doibase 10.1021/nn304277n [ * * , ( ) ] link:\doibase 10.1021/jp3123709 [ * * , ( ) ] link:\doibase 10.1126/science.1089171 [ * * , ( ) ] link:\doibase 10.1021/nl049681c [ * * , ( ) ] link:\doibase 10.1021/nl0486160 [ * * , ( ) ] http://scitation.aip.org/content/aip/journal/jcp/120/11/10.1063/1.1647518 [ * * , ( ) ] link:\doibase 10.1021/cr200061k [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms1806 [ * * , ( ) ] link:\doibase 10.1021/nl4014887 [ * * , ( ) ] link:\doibase 10.1021/jp8083869 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0009-2614(84)85300-2 [ * * , ( ) ] http://scitation.aip.org/content/aip/journal/jcp/83/7/10.1063/1.449120 [ * * , ( ) ] link:\doibase 10.1126/science.1102992 [ * * , ( ) ] link:\doibase 10.1007/s11468 - 007 - 9028 - 9 [ * * , ( ) ] link:\doibase 10.1021/jp062665e [ * * , ( ) ] link:\doibase 10.1021/nl801480 m [ * * , ( ) ] http://link.aps.org/doi/10.1103/physreva.79.053816 [ * * , ( ) ] link:\doibase 10.1021/jp806536w [ * * , ( ) ] link:\doibase 10.1007/s11468 - 010 - 9151-x [ * * , ( ) ] link:\doibase 10.1021/nl101876f [ * * , ( ) ] link:\doibase 10.1364/oe.18.006516 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.cplett.2011.03.088 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physreva.84.053824 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.83.085427 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.86.155305 [ * * , ( ) ] link:\doibase 10.1021/jp309660s [ * * , ( ) ] link:\doibase 10.1021/jp300916a [ * * , ( ) ] link:\doibase 10.1039/c2cp44010e [ * * , ( ) ] http://link.aps.org/doi/10.1103/physreva.89.063817 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1134/s0030400x15010154 [ * * , ( ) ] http://stacks.iop.org/2040-8986/16/i=11/a=114015 [ * * , ( ) ] link:\doibase 10.1021/jp501144s [ * * , ( ) ] http://dx.doi.org/10.1038/nature01937 [ * * , ( ) ] http://dx.doi.org/10.1038/nmat852 [ * * , ( ) ] link:\doibase 10.1364/josaa.21.002442 [ * * , ( ) ] http://dx.doi.org/10.1038/nphoton.2008.131 [ * * , ( ) ] link:\doibase 10.1021/nl901314u [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms1315 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3604014 [ * * , ( ) , http://dx.doi.org/10.1063/1.3604014 ] http://link.aps.org/doi/10.1103/physrevb.84.161304 [ * * , ( ) ] link:\doibase 10.1021/nn3027112 [ * * , ( ) ] link:\doibase 10.1021/nl402358h [ * * , ( ) ] @noop * * ( ) http://dx.doi.org/10.1038/ncomms8334 [ * * ( ) ] http://scitation.aip.org/content/aip/journal/jcp/131/1/10.1063/1.3167407 [ * * , ( ) ] http://scitation.aip.org/content/aip/journal/jcp/134/8/10.1063/1.3541820 [ * * , ( ) ] link:\doibase 10.1103/physreva.84.043802 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physreva.11.1507 [ * * , ( ) ] link:\doibase 10.1103/physreva.47.1247 [ * * , ( ) ] link:\doibase 10.1364/josaa.12.001974 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physreva.59.2342 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physreva.66.063418 [ * * , ( ) ] link:\doibase 10.1109/jstqe.2004.836023 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physreva.78.013806 [ * * , ( ) ] http://jlt.osa.org/abstract.cfm?uri=jlt-27-20-4530 [ * * , ( ) ] link:\doibase 10.1021/nn4054528 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4893967 [ * * , ( ) ] link:\doibase 10.1021/acsphotonics.5b00146 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.92.035433 [ * * , ( ) ] http://stacks.iop.org/0034-4885/78/i=1/a=013901 [ * * , ( ) ] http://stacks.iop.org/1367-2630/15/i=11/a=113014 [ * * , ( ) ] link:\doibase 10.1021/nl502165y [ * * , ( ) ] link:\doibase 10.1063/1.4767653 [ * * , ( ) , 10.1063/1.4767653 ] link:\doibase 10.1117/1.3001731 [ * * , ( ) ] link:\doibase 10.1103/physrevb.68.045415 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physreva.91.043835 [ * * , ( ) ] @noop _ _ , ed .
( , ) \doibase 10.1002/1098 - 2760(20001205)27:5<334::aid - mop14>3.0.co;2-a [ * * , ( ) ] link:\doibase 10.1021/la0513353 [ * * , ( ) ] | we study the transfer of electromagnetic energy across a subwavelength gap separating two co - axial metal nanorodes .
the absence of spacer in the gap separating the rods the system exhibits the strong coupling between longitudinal plasmons in the two rods .
the nature and magnitude of this coupling is studied by varying various geometrical parameters .
when the length of one rod is varied this mode spectrum exhibits the familiar anti - crossing behavior that depends on the coupling strength determined by the gap width . as a function of frequency
the transmission is dominated by a splitted longitudinal plasmon peak .
the two hybrid modes are the dipole - like `` bonding '' mode characterized by a peak intensity in the gap , and a quadrupole - like `` antibonding '' mode whose amplitude vanishes at the gap center . when off - resonant @xmath0level emitters are placed in the gap , almost no effect on the frequency dependent transmission
is observed .
in contrast , when the molecular system is resonant with the plasmonic lineshape , the transmission is strongly modified , showing characteristics of strong exciton - plasmon coupling , modifying mostly the transmission near the lower frequency `` bonding '' plasmon mode .
the presence of resonant molecules in the gap affects not only the molecule - field interaction but also the spatial distribution of the field intensity and the electromagnetic energy flux across the junction . |
the material in this paper can be motivated both by abstract questions in linear algebra about the spectra of sets of matrices , and by the study of asymptotic behaviour in dynamical systems .
the connection is quite natural : given a sufficiently smooth dynamical system , the structure or spectra of the jacobian matrices associated with the system may determine certain behaviours , for example the possibility of various local bifurcations , and more generally the possibility of multiple steady states , oscillations or chaos . given a set @xmath1 and a map @xmath2 , define @xmath3 we may ask , for example : 1 .
`` is @xmath4 ? '' 2 .
`` does @xmath5 intersect the imaginary axis ? '' 3 .
`` does the nonreal part of @xmath6 intersect both left and right open half - planes of @xmath7 ? '' and so forth .
the best - known special case is where @xmath8 is a `` qualitative class '' , namely it consists of all matrices with some sign pattern , in which case q1 reduces to the well - studied _ combinatorial _ problem of characterising sign nonsingular matrices @xcite . by an easy argument , given convex
open @xmath9 , and a @xmath10 function @xmath11 , sign nonsingularity of the jacobian matrices @xmath12 in fact implies injectivity of @xmath13 @xcite .
apart from such applications , exploration of the zero patterns of sign nonsingular matrices has led to a rich combinatorial theory @xcite . if @xmath8 arises as the jacobian matrices of some family of vector fields , a negative answer to either q2 or q3 implies that this family does not admit hopf bifurcation . in this paper
it will be q3 which is of primary interest .
the philosophical approach taken is quite analogous to that taken in the study of sign nonsingularity . as a general principle , when posing some question about a set of matrices @xmath14 , one hopes to associate with elements of @xmath14 discrete objects such as graphs which are constant or vary little over @xmath14 , and then reduce the question to some finite computation on these objects .
the most obvious example is again when @xmath14 is a qualitative class , which can naturally be associated with a signed digraph , or a signed bipartite graph .
however , more generally , products of matrices which have constant sign pattern ( or some of which are constant ) also arise naturally in applications . to see why it is worth going beyond qualitative classes ,
we introduce the notion of an `` interaction network '' as described in @xcite .
consider a system consisting of @xmath15 species with values ( amounts , concentrations , populations , etc . )
define @xmath17 to be the state of the system .
suppose that amongst these species there are @xmath18 interactions , each involving some nonempty subset of the species , and occurring at rates @xmath19 ( @xmath20 ) dependent on the state @xmath21 , but independent of time .
define the rate function @xmath22 by @xmath23 .
finally , for @xmath24 , let @xmath25 describe the total rate of change of species @xmath26 as a function of the interaction rates .
the evolution of the system is then given by : @xmath27 or more briefly , defining @xmath28 by @xmath29 , @xmath30 assume that @xmath13 and @xmath31 are @xmath10 functions , so that by the chain rule , @xmath32 for our purposes here , the important point is that the right hand side of an interaction network ( [ genprodform1 ] ) is defined as a composition of functions , and hence its jacobian matrices have a natural factorisation ( [ decomp ] ) .
this factorisation was the starting point for quite general combinatorial approaches to questions of injectivity and multistationarity in interaction networks in @xcite , and the treatment here relies heavily on the ideas first presented there .
( [ genprodform1 ] ) is not restrictive since an arbitrary system of autonomous first order differential equations can be represented in this way by choosing , for example , one of @xmath31 or @xmath13 to be the identity function . thus casting a system as an interaction network
can not be done uniquely , and can be seen as a formalism for studying certain questions about the system , rather than a categorisation of the system .
however the usefulness of this construction is most apparent when @xmath13 and @xmath31 are defined by natural physical constraints as we shall see below .
models of systems of chemical reactions , termed `` chemical reaction networks '' or crns , naturally take the interaction network form : in fact ( [ genprodform1 ] ) reduces to @xmath33 where @xmath34 is now a constant @xmath35 matrix , termed the `` stoichiometric matrix '' of the system , and @xmath36 is the vector of reaction rates . as the state variables are chemical concentrations , the natural state space is the nonnegative orthant in @xmath37 , i.e. @xmath38 . the jacobian matrices become @xmath39 moreover , under mild physical assumptions , the sign pattern of @xmath40 may be related to that of @xmath34 . for example , if @xmath41 , no chemical occurs on both sides of any reaction , and all chemical reactions are reversible , then we expect @xmath42 to have the same sign as @xmath43 .
this will be discussed further below .
the results will be developed as follows . in section [ sec : prelim ] a considerable volume of preliminary material is gathered for subsequent use .
this includes definitions , notation and known results , along with a few new but relatively straightforward lemmas .
section [ sec : c2decomp ] constructs an object termed the `` dsr@xmath44}$ ] graph '' , which is the central tool required for everything to follow .
this is followed , in section [ sec : results1 ] by a series of claims which can be made immediately from observation of the dsr@xmath44}$ ] graph combined with the results in section [ sec : prelim ] , and without the need for deeper theory .
section [ sec : theory ] contains technical results on the dsr@xmath44}$ ] graph which allow , in section [ sec : final ] some nontrivial extensions of the results of earlier sections .
section [ sec : final ] also describes limitations of the theory developed , and suggests the way forward .
examples are interspersed throughout , and we return to examples to note how the analysis becomes more sophisticated or more rapid as the theoretical tools are developed .
similarly we sometimes prove results which later become corollaries of deeper results ; although this adds to the length of the paper , it is intended to make the material more transparent .
remarks linking results here to related published work , and also to the study of crns , are interspersed throughout .
the frequent reference to crns reflects the fact that although these systems are only one possible application for the theory , they were an important motivator for this work .
preliminaries[sec : prelim ] this paper is concerned with sets of real matrices .
these can be naturally identified with subsets of euclidean space , thus inheriting topological properties such as openness , closedness and connectedness .
we begin with the most fundamental of these sets : a matrix @xmath45 determines the * qualitative class * @xmath46 consisting of all matrices with the same sign pattern as @xmath47 , i.e. , @xmath48 if and only if @xmath49 ; @xmath50 ; and @xmath51 . the closure of @xmath52 will be written @xmath53 , while the closure of a more general @xmath54 will be written @xmath55 .
a square matrix @xmath47 is * sign nonsingular * if all matrices in @xmath52 are nonsingular and * sign singular * if all matrices in @xmath52 are singular @xcite .
[ rem : crnjacobian ] * jacobian matrices of crns . * returning to ( [ genprodformcrn ] ) and the subsequent discussion , the notation developed above allows us to abbreviate the condition `` @xmath42 has the same sign as @xmath43 for all @xmath56 '' as @xmath57 .
more generally , for systems of reactions where no chemical occurs on both sides of any reaction , we expect the condition @xmath58 to hold on the entire nonnegative orthant , including the boundary . even when this does not hold ( namely when some chemical occurs on both sides of some reaction ) , provided every reversible reaction is treated as a pair of irreversible reactions , we still expect @xmath40 to belong to the closure of a qualitative class which can be inferred from the network structure .
thus quite generally , the jacobian matrices of a crn take the form of a constant matrix times a matrix belonging to some qualitative class .
given an @xmath35 matrix @xmath47 , and nonempty @xmath59 , @xmath60 , define @xmath61 to be the submatrix of @xmath47 obtained by choosing rows of @xmath47 from @xmath62 and columns from @xmath63 , and ( provided @xmath64 ) @xmath65 $ ] to be the corresponding minor of @xmath47 .
@xmath66 $ ] will be an abbreviation for the principal minor @xmath67 $ ] .
given an @xmath68 matrix @xmath47 and some permutation @xmath69 of @xmath70 denote by @xmath71 the sign of this permutation , and by @xmath72 the term @xmath73 in @xmath74 , so that @xmath75 .
consider an @xmath76 matrix @xmath77 and an @xmath78 matrix @xmath79 for arbitrary positive integers @xmath80 .
the cauchy - binet formula ( @xcite for example ) tells us that for any nonempty @xmath59 , @xmath60 with @xmath64 : @xmath81 = \sum_{\substack{\gamma \subseteq \{1 , \ldots , m\}\\ |\gamma| = |\alpha|}}a[\alpha|\gamma]b[\gamma|\beta].\ ] ] this formula for the minors of a product is central to many of the matrix and graph - theoretic results developed and cited below .
this formula also has a briefer statement in terms of compound matrices mentioned later . given a square matrix @xmath47 , define @xmath82 to be the multiset of eigenvalues of @xmath47 . denote by @xmath83 ( resp .
@xmath84 ) the open ( resp
. closed ) left half - plane of @xmath7 with @xmath85 and @xmath86 similarly defined .
a square matrix @xmath47 whose spectrum lies entirely in @xmath85 ( resp .
@xmath86 ) is * positive stable * ( resp . *
positive semistable * ) .
@xmath47 is * hurwitz * if all of its eigenvalues lie in @xmath83 , which occurs iff @xmath87 is positive stable .
* @xmath88-matrices * are square matrices all of whose principal minors are positive .
* @xmath89-matrices * are matrices in the closure of the @xmath88-matrices , i.e. , all of whose principal minors are nonnegative . here
a @xmath88-matrix ( or @xmath89-matrix ) will refer to a real matrix . of particular importance will be the following restrictions on the spectra of @xmath89-matrices and @xmath88-matrices proved in @xcite : [ p0eigenvals ] a complex number @xmath90 is an eigenvalue of an @xmath68 @xmath89-matrix if and only if @xmath91 or latexmath:[\[\label{eg : p0wedge } of an @xmath68 @xmath88-matrix if and only if @xmath93 and @xmath94 this implies , in particular that no @xmath89-matrix can have a negative real eigenvalue .
obviously , if @xmath95 is a @xmath89-matrix , then @xmath96 can not have a positive real eigenvalue .
we will also need the following well - known fact about the spectrum of a product of matrices : [ spectranposelemma ] given @xmath97 , @xmath98 .
we define @xmath99 as the @xmath100th exterior power of @xmath37 .
in particular , @xmath101 , @xmath102 , and for @xmath103 , the elements of @xmath99 are finite formal linear combinations of elements of the form @xmath104 where @xmath105 , and `` @xmath106 '' is the wedge - product ( see @xcite for example ) .
let @xmath47 denote a linear transformation from @xmath107 to @xmath37 , and also its matrix representation in some basis .
@xmath47 determines transformations from @xmath108 to @xmath99 in two important ways @xcite : 1 .
@xmath109 is defined by : @xmath110 in this case the map @xmath111 is termed the @xmath100th exterior power of @xmath47 , or the @xmath100th multiplicative compound of @xmath47 . choosing some bases for @xmath108 and @xmath99 ,
the resulting @xmath112 matrix will also be denoted @xmath111 .
if @xmath113 , @xmath114}:\lambda^k\mathbb{r}^n \to \lambda^k\mathbb{r}^n$ ] is defined by : @xmath115}(u_1 \wedge \cdots \wedge u_k ) = \sum_{i=1}^ku_1 \wedge \cdots ( mu_i ) \wedge \cdots \wedge u_k\,.\ ] ] in this case the map @xmath114}$ ] is termed the @xmath100th additive compound of @xmath47 . choosing a basis for @xmath99 , the resulting matrix is @xmath116 , and will also be denoted @xmath114}$ ] .
note that @xmath117 } = -(m^{[2]})$ ] so we can write without ambiguity @xmath118}$ ] .
multiplicative and additive compound matrices can easily be shown to satisfy the relationships @xmath119 ( provided @xmath77 and @xmath79 are of dimensions such that @xmath120 makes sense ) and @xmath121 } = a^{[k]}+ b^{[k]}$ ] ( provided @xmath77 and @xmath79 are square and of the same dimensions ) .
@xmath119 is in fact a restatement of the cauchy - binet formula in terms of multiplicative compounds .
when referring to a compound matrix we assume without comment that the dimensions are such that the matrix exists .
for example , if @xmath122}$ ] mentioned , it is to be assumed that @xmath47 is a square matrix of dimension at least @xmath123 .
compound matrices of a square matrix @xmath124 appear naturally in the study of various questions on the stability of differential equations ( see @xcite for example )
. we will refer to some of this theory , but most important here are basic spectral properties of compound matrices : for @xmath125 , the @xmath126 eigenvalues of @xmath111 are precisely _ products _ of all sets of @xmath100 distinct elements of @xmath82 , while the @xmath126 eigenvalues of @xmath114}$ ] are the _ sums _ of all sets of @xmath100 distinct elements of @xmath82 .
here it is second additive compound matrices which will be of most interest : if @xmath127 then @xmath128})=\{\lambda_i+\lambda_j:1\le i
< j\le n\}.\ ] ] second additive compound matrices are sometimes presented by means of a _ bialternate product _ @xcite .
this reference also demonstrates that second additive compounds are natural objects to consider when studying hopf bifurcation . for us ,
the following two observations are important : [ j2nonsingular ] if @xmath47 is a real square matrix and @xmath122}$ ] is nonsingular , then @xmath129 does not intersect the imaginary axis .
if @xmath129 intersects the imaginary axis , then @xmath47 has a pair of eigenvalues @xmath130 where @xmath131 .
these sum to zero and hence @xmath122}$ ] is singular .
[ j2p0 ] if @xmath47 is a real square matrix and @xmath122}$ ] is a @xmath89-matrix , then the nonreal part of @xmath132 ( i.e. @xmath133 ) does not intersect @xmath83 .
if @xmath118}$ ] is a @xmath89-matrix , then the nonreal part of @xmath132 does not intersect @xmath85 .
if the nonreal part of @xmath132 intersects @xmath83 , then @xmath47 has a pair of eigenvalues @xmath134 where @xmath135 . these sum to @xmath136 and hence @xmath122}$ ]
has a negative real eigenvalue , forbidden by lemma [ p0eigenvals ] .
the second statement follows immediately since the spectrum of @xmath118}$ ] is simply the reflection in the imaginary axis of that of @xmath122}$ ] .
[ rem : adddiag ] if @xmath122}$ ] is a @xmath89-matrix , then in fact lemma [ j2p0 ] extends immediately as follows : given any nonnegative diagonal matrix @xmath137 , the nonreal part of @xmath138 does not intersect @xmath83 .
this follows because ( i ) @xmath139 } = m^{[2 ] } + d^{[2]}$ ] , ( ii ) @xmath140}$ ] is itself a nonnegative diagonal matrix , and ( iii ) the set of @xmath89-matrices is closed under the addition of nonnegative diagonal matrices .
this fact is of importance in applications to crns , where diagonal terms represent arbitrary degradation or outflow of chemicals ( see @xcite ) .
this will be commented on further below .
we also have the following two lemmas about stability : [ lemposstable0 ] consider a square matrix @xmath47 . if both @xmath47 and @xmath122}$ ] are @xmath89-matrices ( resp .
@xmath88-matrices ) , then @xmath47 is positive semistable ( resp .
positive stable ) . since @xmath47 is a @xmath89-matrix ( resp .
@xmath88-matrix ) by lemma [ p0eigenvals ] the real part of @xmath132 does not intersect @xmath83 ( resp .
@xmath84 ) . since @xmath122}$ ] is a @xmath89-matrix ( resp .
@xmath88-matrix ) , by lemma [ j2p0 ] , the nonreal part of @xmath132 does not intersect @xmath83 ( resp .
@xmath84 ) . combining these two facts
, @xmath47 is positive semistable ( resp .
positive stable ) .
[ rem : adddiaga ] continuing from remark [ rem : adddiag ] , if the conditions of lemma [ lemposstable0 ] hold , namely both @xmath47 and @xmath122}$ ] are @xmath89-matrices ( resp .
@xmath88-matrices ) , then given any nonnegative diagonal matrix @xmath137 , @xmath141 is positive semistable ( resp .
positive stable ) .
[ lemposstable ] consider a set @xmath142 which is path - connected .
suppose that for each @xmath143 , both @xmath47 and @xmath122}$ ] are nonsingular and that @xmath144 includes a positive stable ( resp .
hurwitz ) matrix .
then @xmath144 consists entirely of positive stable ( resp .
hurwitz ) matrices .
let @xmath145 be a positive stable ( resp .
hurwitz ) matrix which exists by assumption .
suppose @xmath144 includes a matrix @xmath146 which fails to be positive stable ( resp .
hurwitz ) .
since @xmath144 is path - connected , and eigenvalues of a matrix depend continuously on the entries of a matrix , there must lie , on any path in @xmath144 connecting @xmath147 and @xmath146 , a matrix @xmath143 with some eigenvalue on the imaginary axis .
however , since both @xmath47 and @xmath122}$ ] are nonsingular , by lemma [ j2nonsingular ] , @xmath47 has no eigenvalues on the imaginary axis .
the stability criterion in lemma [ lemposstable ] is closely related to that of li and wang in @xcite , and indeed , it is possible to use the spectral properties of compound matrices to develop other stability criteria for sets of matrices which can be more useful in practice than attempting directly to check the routh - hurwitz conditions .
a one - parameter family of vector fields @xmath148 generically undergoes a poincar - andronov - hopf bifurcation ( from here on abbreviated to `` hopf bifurcation '' ) if there exists some @xmath149 such that @xmath150 , @xmath151 has a pair of imaginary eigenvalues and moreover these eigenvalues cross the imaginary axis as @xmath152 varies through @xmath153 ( @xcite for example ) .
thus vector fields can be shown to _
forbid _ hopf bifurcation either if the nonzero spectrum of each jacobian matrix avoids the imaginary axis , or the nonreal spectrum of each jacobian matrix avoids @xmath83 or @xmath85 .
let @xmath154 and @xmath155 be the real and imaginary axes in @xmath7 .
as indicated earlier , we are particularly interested in matrix products . in particular , when studying hopf bifurcation , we are led to the following questions : 1 . given @xmath156 and @xmath157 ,
is @xmath158 ?
2 . given @xmath156 and @xmath157 ,
is @xmath159 ?
since under certain assumptions , the jacobian matrix of a crn is of the form @xmath160 where @xmath161 ( see remark [ rem : crnjacobian ] ) , when analysing the behaviour of crns we are led to the questions : 1 . given @xmath162 , is @xmath163 ?
2 . given @xmath162 , is @xmath164 ? in this paper
we begin the process of developing tools which provide answers to these questions .
[ rem : chemhopf ] from lemma [ j2p0 ] , the reader may already guess that our main tools for ruling out hopf bifurcations here will involve proving that @xmath165}$ ] or @xmath166}$ ] is a @xmath89-matrix for each allowed jacobian matrix of some system . continuing the theme of remark [ rem : adddiag ] ,
if hopf bifurcation is ruled out for a chemical system with jacobian matrices @xmath14 because @xmath166}$ ] is a @xmath89-matrix for all @xmath167 , then it is in fact ruled out even if arbitrary degradation or outflow reactions are added .
( analogous statements follow replacing `` @xmath166}$ ] '' with `` @xmath95 '' and `` hopf bifurcation '' with `` saddle - node bifurcation '' , but this is well - studied @xcite and not the theme of this paper . ) as indicated earlier , second additive compounds arise naturally in the study of hopf bifurcations , a theme discussed thoroughly in @xcite .
in that work , the focus lies on devising continuation methods to detect hopf bifurcations , with @xmath165}$ ] occurring in algebraic systems which augment the equilibrium condition . on the other hand ,
our goal here is to present conditions which rely on combinatorial features of @xmath165}$ ] to forbid hopf bifurcation .
these conditions translate into graphical requirements involving * dsr graphs * , which are discussed next .
a class of generalised graphs , sometimes termed sr graphs and dsr graphs , have become useful tools in the study of properties of interaction networks , such as multistationarity and stability @xcite .
dsr graphs for general interaction networks were constructed originally in @xcite .
the definition presented here involves some minor modifications of that in @xcite for ease of presentation and maximum generality .
given @xmath162 and @xmath168 , the dsr graph @xmath169 is defined as follows : we begin with a signed bipartite digraph on @xmath170 vertices , @xmath171 , termed s - vertices , and @xmath172 , termed r - vertices , with arc @xmath173 if and only if @xmath174 , and arc @xmath175 if and only if @xmath176 . the arc @xmath173 is said to have r - to - s orientation and is given the sign of @xmath177 ; the arc @xmath175 is said to have s - to - r orientation and is given the sign of @xmath178 .
in addition : 1 .
a pair of antiparallel edges of the same sign are treated as a single undirected edge , so that each edge of @xmath169 either has s - to - r orientation or r - to - s orientation or both , in which case we say that the edge is undirected .
when discussing the degree of a vertex , an undirected edge incident into the vertex contributes exactly @xmath179 to the degree .
it is useful to give the ( directed or undirected ) edge @xmath173 associated with a nonzero entry @xmath177 the label @xmath180 .
we refer to this label of an edge @xmath181 as @xmath182 .
if @xmath181 has only s - to - r orientation ( i.e. it is associated only with an entry from @xmath79 ) , then we formally define @xmath183 .
this choice of labelling convention springs from the fact that we can generally consider @xmath77 to be constant if the system being studied is a crn ( see remark [ rem : crnjacobian ] ) .
[ rem : dsrterminology ] what is here termed @xmath169 would be @xmath184 in the terminology of @xcite .
further , the construction here is presented for a single pair of matrices rather than directly for sets of matrices ; note however that if @xmath77 is fixed and @xmath79 varies over a qualitative class , then the dsr graph remains constant . on the other hand , if @xmath77 varies ( within a qualitative class ) , then varying edge labels can be replaced with the formal label @xmath185 .
[ figdsr ] below we illustrate the dsr graph @xmath186 associated with two matrices @xmath77 and @xmath79 . negative edges ( corresponding to negative entries in the matrices ) are shown as dashed lines , while positive edges are bold lines , a convention which will be followed throughout this paper . to make the construction more transparent ,
an intermediate stage is shown where separate signed digraphs @xmath187 and @xmath188 are constructed from @xmath77 and from @xmath79 : @xmath186 can be regarded as the `` superposition '' of these two digraphs .
edge - labels are associated only with @xmath77 : an edge imported only from @xmath79 is labelled @xmath185 .
[ scale=0.54 , place/.style = circle , draw = blue!50,fill = blue!5,thick , inner sep=0pt , minimum size=5.5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.5 mm , transition/.style = rectangle , draw = black!50,fill = black!5,thick , inner sep=0pt , minimum size=5 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; at ( -4.1,7.5 ) @xmath189 ; ( s1 ) at ( 1,9 ) @xmath190 ; ( r2 ) at ( 4,9 ) @xmath191 ; ( s2 ) at ( 7,9 ) @xmath192 ; ( r1 ) at ( 1,6 ) @xmath193 ; ( s3 ) at ( 4,6 ) @xmath194 ; ( r2 ) ( s1 ) ; ( r2 )
( s2 ) ; ( r2 ) ( s3 ) ; ( r1 ) ( s1 ) ; ( r1 ) ( s3 ) ; at ( 2.5 , 9.3 ) @xmath195 ; at ( 2.5 , 6.3 ) @xmath196 ; at ( 1.3 , 7.5 ) @xmath197 ; at ( 4.3 , 7.5 ) @xmath197 ; at ( 5.5 , 9.3 ) @xmath198 ; at ( -4.1,2.5 ) @xmath199 ; ( s1 ) at ( 1,4 ) @xmath190 ; ( r2 ) at ( 4,4 ) @xmath191 ; ( s2 ) at ( 7,4 ) @xmath192 ; ( r1 ) at ( 1,1 ) @xmath193 ; ( s3 ) at ( 4,1 ) @xmath194 ; ( r2 ) ( s1 ) ; ( r2 ) ( s2 ) ; ( r1 ) ( s1 ) ; ( r1 ) ( s3 ) ; ( 8.5 , 0.5 ) .. controls ( 9,0.5 ) and ( 9 , 0.5 ) .. ( 9,2.5 ) ; ( 9 , 2.5 ) .. controls ( 9,5 ) and ( 9 , 5 ) .. ( 9.5,5 ) ; ( 8.5 , 9.5 ) .. controls ( 9,9.5 ) and ( 9 , 9.5 ) .. ( 9,7.5 ) ; ( 9 , 7.5 ) .. controls ( 9,5 ) and ( 9 , 5 ) .. ( 9.5,5 ) ; ( s1 ) at ( 11,6.5 ) @xmath190 ; ( r2 ) at ( 14,6.5 ) @xmath191 ; ( s2 ) at ( 17,6.5 ) @xmath192 ; ( r1 ) at ( 11,3.5 ) @xmath193 ; ( s3 ) at ( 14,3.5 ) @xmath194 ; ( r2 ) ( s1 ) ; ( r2 ) ( s2 ) ; ( r2 ) ( s3 ) ; ( r1 ) ( s1 ) ; ( r1 ) .. controls ( 12.2,3.9 ) and ( 12.8 , 3.9 ) .. ( s3 ) ; ( r1 ) .. controls ( 12.2,3.1 ) and ( 12.8 , 3.1 ) .. ( s3 ) ; at ( 12.5 , 6.8 ) @xmath195 ; at ( 12.5 , 4.2 ) @xmath196 ; at ( 12.5 , 2.8 ) @xmath200 ; at ( 11.3 , 5 ) @xmath197 ; at ( 14.3 , 5 ) @xmath197 ; at ( 15.5 , 6.8 ) @xmath198 ; at ( 14 , 1 ) @xmath186 ; [ remsignpat ] a special case relevant to the study of crns is when the matrices @xmath201 satisfy @xmath202 . in this case
all edges of @xmath186 are undirected , and in fact @xmath203 .
this graph can also be referred to as the * sr graph * corresponding to the matrix @xmath77 . to present the main results about dsr graphs ,
some further definitions are needed .
a * walk * @xmath204 in a digraph is an alternating sequence of vertices and edges , beginning and ending with a vertex , and where each edge in @xmath204 is preceded by its start - point and followed by its end - point , each vertex ( except the first ) is preceded by an edge incident into it , and each vertex ( except the last ) is followed by an edge incident out of it .
here we allow a walk @xmath204 to repeat both vertices and edges , and its * length * @xmath205 is the number of edges in @xmath204 , counted with repetition .
if the first and last vertex are the same , the walk is called * closed*. in what follows we may refer to walks by their sequence of edges , or vertices , or both and we will say that a walk is empty and denote it as @xmath206 if it contains no edges ( an empty walk may include a single vertex ) . when used without qualification , `` walk '' will mean a nonempty walk .
the * sign * of a walk is defined as the product of signs of the multiset of edges in the walk .
the empty walk is formally given the sign @xmath207 . a closed nonempty walk which does not repeat vertices ( except in the trivial sense that the first and last vertices are the same ) , is called a * cycle*. clearly a cycle has no repeated edges . a dsr graph with no cycles will be termed acyclic : note that @xmath208 is considered a cycle in a dsr graph if and only if @xmath209 and @xmath210 are directed edges with different signs and so are not treated as a single edge . as dsr graphs are bipartite objects , each closed walk @xmath204 has even length .
any walk of even length can be given a parity @xmath211 defined as : @xmath212 if @xmath213 , then @xmath204 is termed * even * , and if @xmath214 , then @xmath204 is termed * odd*. a cycle @xmath215 is termed an * e - cycle * if @xmath216 , and an * o - cycle * if @xmath217 .
a closed walk @xmath218 is an * s - walk * if each edge @xmath181 in @xmath204 has an associated label @xmath219 and @xmath220 if @xmath204 is a cycle which is an s - walk , it is termed an * s - cycle*. a cycle in a dsr graph which includes some edge with only s - to - r orientation or only r - to - s orientation has a unique orientation ; otherwise the cycle has two possible orientations .
two oriented cycles in a dsr graph are compatibly oriented , if each induces the same orientation on every edge in their intersection .
two cycles ( possibly unoriented ) are compatibly oriented if there is an orientation for each so that this requirement is fulfilled . in a dsr graph , two cycles have * odd intersection * if they are compatibly oriented and each component of their intersection has odd length . note that odd intersection was termed `` s - to - r intersection '' in @xcite .
[ remoddintersect ] a necessary condition for the existence of two cycles , say @xmath215 and @xmath137 , with odd intersection in a dsr graph is the existence of at least one s - vertex of degree @xmath221 or more , and of at least one r - vertex of degree @xmath221 or more .
this follows since the vertices at the ends of a component of the intersection of @xmath215 and @xmath137 of odd length must have degree at least @xmath221 , and one of these must be an s - vertex and one an r - vertex .
a matching in a dsr graph is a set of edges without common vertices .
an s - to - r matching is a matching all of whose edges have s - to - r orientation , and an r - to - s matching is similarly defined .
a matching which covers all vertices is perfect
. we will refer to a dsr graph @xmath222 as * odd * if @xmath222 has no e - cycles
. we will refer to it as * odd@xmath223 * if all e - cycles in @xmath222 are s - cycles , and no two e - cycles have odd intersection .
it will be called * steady * if all cycles in @xmath222 are s - cycles .
we summarise several important results connecting dsr graphs and @xmath89-matrices in a single theorem .
[ mainp0thm ] define the following conditions on matrices @xmath162 and @xmath168 and associated graphs : * @xmath169 is odd .
* @xmath169 is odd@xmath223 .
* @xmath224 is a @xmath89-matrix for all @xmath225 , @xmath226 .
* @xmath227 is a @xmath89-matrix for all @xmath226 .
* @xmath228 is odd .
* @xmath228 is odd@xmath223 . *
every square submatrix of @xmath77 is either sign nonsingular or sign singular . *
every square submatrix of @xmath77 is either sign nonsingular or singular .
* @xmath229 is a @xmath89-matrix for all @xmath225 , @xmath230 .
* @xmath231 is a @xmath89-matrix for all @xmath230 .
the following implications hold : @xmath232 @xmath233 the implications @xmath234 , @xmath235 , @xmath236 , @xmath237 and @xmath238 follow by definition , while the other implications are results in @xcite , or are immediate consequences of these results . note that @xmath239 and @xmath240 are just immediate corollaries of @xmath241 and @xmath242 respectively .
[ rem : p0proof ] although some of the proofs of results in theorem [ mainp0thm ] are lengthy , the spirit is worth outlining . consider matrices @xmath243 and dsr graph @xmath169 .
since the results are about the product @xmath120 , they begin with the cauchy - binet formula ( [ eqcb0 ] ) , which provides an expression for each minor of @xmath120 as a sum of products of minors of @xmath77 and @xmath79 .
the transition from matrices to dsr graphs is via the association of terms in minors with matchings in dsr graphs . given appropriate sets @xmath244 a term in @xmath245 $ ] corresponds to an r - to - s matching in @xmath169 which is perfect on the vertices @xmath246 , while a term in @xmath247 $ ] corresponds to an s - to - r matching which is perfect on the same vertex set .
the union of these matchings is a subgraph of @xmath169 consisting of ( disjoint ) cycles and isolated , undirected , edges .
conclusions about the signs and values of terms in the minors follow from examination of these subgraphs . alongside theorem [ mainp0thm ]
, we will need a few further results specifically about steady dsr graphs , which were not considered in the references above , but become important in the discussion here .
[ lem : swalks ] let @xmath222 be a steady dsr graph . then any closed walk on @xmath222 is an s - walk .
let @xmath204 be a closed walk on @xmath222 .
certainly each edge of @xmath204 which occurs in a cycle of @xmath222 has finite label , since all cycles are s - cycles .
further , each edge of @xmath204 not occurring in a cycle of @xmath222 must be undirected , and consequently have a finite label : traversing arc @xmath209 and returning subsequently to @xmath248 otherwise implies the existence of a cycle which includes @xmath209 .
thus all edges in @xmath204 have finite labels .
we show that @xmath249 if no vertex is repeated in @xmath204 , then @xmath204 is either a cycle , in which case @xmath250 by hypothesis , or a single edge traversed back and forward , in which case again @xmath250 .
if , on the other hand , @xmath204 contains repeated vertices , then the shortest closed subwalk of @xmath251 denoted @xmath252 , is either a cycle or a single repeated edge .
denote by @xmath253 the closed subwalk obtained by removing @xmath252 from @xmath204 , namely if the vertex - list of @xmath204 is @xmath254 and that of @xmath252 is @xmath255 , then for @xmath253 we have @xmath256 .
clearly @xmath257 if the walk @xmath253 does not repeat vertices , we are done .
otherwise , we may repeat this procedure until the remaining walk has no repeated vertices , and the lemma is proved . since an acyclic dsr graph vacuously fulfils the conditions of lemma [ lem : swalks ] ,
an immediate corollary is : [ cor : tree - s ] let @xmath222 be an acyclic dsr graph .
any closed walk on @xmath222 is an s - walk .
[ lem : scyclepmat ] consider some @xmath258 , and some @xmath259 such that for each @xmath143 , @xmath260 is steady . given any @xmath261 , @xmath120 is a @xmath89-matrix .
if the premise of the theorem implies that @xmath120 is a @xmath89-matrix for each @xmath262 , then the result follows for @xmath263 since the set of @xmath89-matrices is closed in @xmath264 .
so consider @xmath265 .
given any nonempty @xmath59 , by the cauchy - binet formula : @xmath266 = \sum_{\substack{\gamma \subseteq \{1 , \ldots , m\}\\ |\gamma| = |\alpha|}}a[\alpha|\gamma]b[\gamma|\alpha].\ ] ] fix @xmath267 , and for brevity define @xmath268 and @xmath269 .
suppose @xmath72 and @xmath270 are two nonzero terms in @xmath271 $ ] corresponding to permutations @xmath272 of @xmath273 .
note that @xmath274 @xmath72 and @xmath270 each correspond to a matching in @xmath228 , and the union of these two matchings defines a set of disjoint cycles corresponding to nontrivial cycles in the permutation @xmath275 and isolated edges corresponding to trivial cycles in @xmath275 ( see remark [ rem : p0proof ] ) .
since @xmath228 is steady , we get @xmath276 ( see the proof of lemma 7 in @xcite for example ) . thus @xmath277
= r(\alpha , \gamma)|m_{(\sigma)}|$ ] where @xmath278 is an integer independent of the values of nonzero entries in @xmath77 . since @xmath279 belongs to the same qualitative class as @xmath77 and all cycles in @xmath280 are s - cycles , we get @xmath281 = r(\alpha , \gamma)|n_{(\sigma)}|$ ] where @xmath282 . thus @xmath283b[\gamma|\alpha ] = [ r(\alpha , \gamma)]^2\,|m_{(\sigma)}|\,\,|n_{(\sigma)}|
\geq 0\,,\ ] ] and the result that @xmath120 is a @xmath89-matrix now follows from ( [ eqcb1 ] ) . compound matrices and the dsr@xmath44}$ ] graph[sec : c2decomp ] the main purpose of this section is as follows : given a square matrix product @xmath120 , we wish to be able to write @xmath284 } = \tilde a \tilde b$ ] where @xmath285 depends only on @xmath77 , and @xmath286 depends only on @xmath79 .
this will allow us to construct an object , the dsr@xmath44}$ ] graph of @xmath120 , whose structure encodes information about @xmath284}$ ] allowing the use of theorem [ mainp0thm ] and subsequent results . to motivate the construction we first present two examples : [ example_motivate ] consider the matrices @xmath287 assume that @xmath288 . here
, as in all examples to follow , we let @xmath289 denote the @xmath26th standard basis vector in @xmath37 and choose the natural basis for @xmath290 consisting of the wedge - products of pairs of vectors @xmath289 , arranged in lexicographic order .
so for @xmath290 , using the basis @xmath291 , we can compute : @xmath292 } = \left(\begin{array}{ccc}a+b+c+d+e & f+g & -f\\d+e & a+b+f+g & c+d\\-b & a+b & c+d+e+f+g\end{array}\right).\ ] ] by computing its @xmath293 principal minors , we can check directly that @xmath165}$ ] is a @xmath89-matrix ( although @xmath96 itself need not be ) .
we will see later that the conclusion that @xmath165}$ ] is a @xmath89-matrix can be reached without symbolic computation , and this example is a special case of a much more general result .
[ example_motivate0 ] consider the matrices @xmath201 defined by : @xmath294 where @xmath295 are all positive real numbers .
@xmath120 is a @xmath296 matrix , so @xmath284}$ ] is a @xmath297 matrix .
( i ) we can check that @xmath120 is a @xmath89-matrix .
( ii ) with the help of a symbolic algebra package such as maxima @xcite , we can compute the @xmath298 principal minors of @xmath284}$ ] , confirm that these are all nonnegative , and conclude that @xmath284}$ ] is a @xmath89-matrix . by lemma [ lemposstable0 ] , ( i ) and ( ii )
imply that @xmath120 is positive semistable .
it is again natural to ask whether this conclusion can be reached simply , via theorem [ mainp0thm ] , and without the use of symbolic algebra .
we will return to this example and show that this is indeed the case .
we will eventually write : @xmath300 } = \overline{\mathbf{l}}^a\underline{\mathbf{l}}^b\ ] ] where the matrix @xmath301 depends on @xmath77 and @xmath302 depends on @xmath79 .
the linear algebraic justification for the factorisation begins by identifying @xmath299 with the set of @xmath68 antisymmetric matrices .
define the matrix @xmath303 by @xmath304 note that @xmath305 .
denote by @xmath306 the subspace of @xmath264 spanned by @xmath307 , i.e. the space of all real @xmath68 antisymmetric matrices . given any @xmath308
define @xmath309 via @xmath310 . identifying @xmath311 with @xmath312 ,
@xmath313 can be identified with the second additive compound @xmath314}$ ] . to be precise , define @xmath315 via @xmath316 .
that this action on basis vectors extends naturally to give a well - defined linear bijection can easily be checked .
we can then confirm that given any square matrix @xmath215 , the following diagram commutes : at ( 1,4 ) @xmath299 ; at ( 4.2,4 ) @xmath299 ; at ( 0.8,1 ) @xmath306 ; at ( 4.2,1 ) @xmath306 ; ( 1.7,4.0 )
( 3.5 , 4.0 ) ; ( 1.8,1.0 )
( 3.2 , 1.0 ) ; ( 1.0 , 3.6 ) ( 1.0 , 1.4 ) ; ( 4.0 , 3.6 ) ( 4.0 , 1.4 ) ; at ( 0.6 , 2.5 ) @xmath317 ; at ( 2.5 , 0.6 ) @xmath318 ; at ( 2.5 , 4.4 ) @xmath314}$ ] ; at ( 4.4 , 2.5 ) @xmath317 ; the calculation is straightforward : @xmath319})(e_i\wedge e_j ) & = & \iota(ce_i \wedge e_j + e_i \wedge ce_j)\\ % & = & \iota(ce_i \wedge e_j ) + \iota(e_i \wedge ce_j ) \\ & = & \iota\left(\sum_kc_{ki}e_k \wedge e_j\right ) + \iota\left(e_i \wedge \sum_kc_{kj}e_k\right)\\ & = & \sum_kc_{ki}\iota(e_k \wedge e_j ) + \sum_kc_{kj}\iota(e_i \wedge e_k)\\ & = & \sum_kc_{ki}x_{kj } + \sum_kc_{kj}x_{ik}\\ & = & \sum_k\left[c_{ki}(e_ke_j^t - e_je_k^t ) + c_{kj}(e_ie_k^t - e_ke_i^t)\right]\\ & = & ce_ie_j^t- e_je_i^tc^t + e_ie_j^tc^t - ce_je_i^t\\ & = & cx_{ij } + x_{ij}c^t = l^cx_{ij } = ( l^c\circ \iota)(e_i\wedge e_j)\,.\end{aligned}\ ] ] the construction is first carried out for rank @xmath179 matrices and extended to arbitrary matrices . given @xmath320 , we can define the linear map @xmath321 via its action on coordinate vectors : @xmath322 noting that if @xmath323 and @xmath324 , then @xmath325 ( where @xmath326 is the kronecker delta ) , a matrix representation of @xmath327 is obtained : @xmath328_{(i , j ) } \quad ( i < j ) \nonumber\\ & = & \sum_{l = k+1}^ns_le_i^tx_{kl}e_j - \sum_{l=1}^{k-1}s_le_i^tx_{lk}e_j \nonumber\\ \label{eq : matrep } & = & \sum_{l = k+1}^ns_l\delta_{ki}\delta_{jl } - \sum_{l=1}^{k-1}s_l\delta_{kj}\delta_{il } = \left\{\begin{array}{cc}s_j & ( k= i)\\-s_i&(k = j)\\0&\mbox{otherwise}\end{array}\right.\end{aligned}\ ] ] for example if @xmath329 , then @xmath330 row and column labels have been shown explicitly to highlight the fact that each column is associated with an entry in @xmath331 , while each row is associated with a _ pair _ of entries .
if @xmath31 is a @xmath332 matrix ( a row vector ) then @xmath333 is similarly defined via its action on basis elements of @xmath306 : @xmath334 with @xmath335 , we obtain the coordinate representation : @xmath336_k = e_k^t(e_iv_j - e_jv_i ) = \delta_{ki}v_j - \delta_{kj}v_i = \left\{\begin{array}{cc}v_j & ( k= i),\\-v_i&(k = j),\\0&\mbox{otherwise.}\end{array}\right.\ ] ] thus @xmath337 .
now consider some rank @xmath179 matrix @xmath338 where @xmath339 , and the associated operator @xmath340 . applying the definitions : @xmath341 in other words , @xmath342 .
thus representations of @xmath327 and @xmath343 above give us a natural representation and factorisation of @xmath344 . more generally , consider any @xmath162 and @xmath168 .
defining @xmath345 to be the @xmath100th column of @xmath77 and @xmath346 to be the @xmath100th row of @xmath79 we have @xmath347 .
we get @xmath348 choosing a basis for @xmath299 we define for any @xmath35 matrix @xmath77 the matrices @xmath349 \quad \mbox{and } \quad \underline{\mathbf{l}}^a = \left[\begin{array}{c}\underline{\mathbf{l}}^{a^1}\\\underline{\mathbf{l}}^{a^2}\\\vdots\\\underline{\mathbf{l}}^{a^n}\end{array}\right]\,.\ ] ] note that @xmath350 . given @xmath351 and @xmath352 we get : @xmath353 } = \overline{\mathbf{l}}^a\underline{\mathbf{l}}^b\,.\ ] ] observe that @xmath301 has dimensions @xmath354 and @xmath302 has dimensions @xmath355 .
[ rem : cb ] when @xmath77 and @xmath79 are matrices with symbolic entries , one approach to computation of any minor of @xmath284}$ ] is by application of the cauchy - binet formula to ( [ eq : j2decomp ] ) . for example assuming that @xmath77 is @xmath35 and @xmath79 is @xmath356 , and defining @xmath357 : @xmath358 } ) = \sum_{\substack{\beta \subseteq \{1 , \ldots , nm\}\\ |\beta| = { n\choose 2}}}\overline{\mathbf{l}}^a[\alpha|\beta]\underline{\mathbf{l}}^b[\beta|\alpha]\,.\ ] ] this observation can be used in conjunction with lemma [ j2nonsingular ] to rule out the possibility of pairs of imaginary eigenvalues of @xmath120 . given @xmath162 and @xmath168 with @xmath359
we define the * dsr@xmath44}$ ] graph * of the product @xmath120 @xmath360}_{a , b } = g_{\overline{\mathbf{l}}^a,\ , \underline{\mathbf{l}}^{b}}\ ] ] to be the dsr graph of the product @xmath361 .
examination of this graph allows us to make statements about @xmath284}$ ] using , for example , theorem [ mainp0thm ] .
the s - vertices of @xmath362}_{a , b}$ ] are the @xmath363 unordered pairs @xmath364 , @xmath365 ; the r - vertices of @xmath362}_{a , b}$ ] are the @xmath366 pairs @xmath367 . for notational convenience
we denote s - vertices @xmath368 by @xmath369 , or @xmath370 ; r - vertices @xmath371 by @xmath372 ; and an edge between them of unknown direction and sign by @xmath373 from the definitions , a necessary condition for the edge @xmath374 to exist is @xmath375 .
[ remsignpat1 ] continuing from remark [ remsignpat ] , it is clear that given a matrix @xmath77 and some @xmath202 , @xmath362}_{a , b } = g^{[2]}_{a , a^t}$ ] .
it is also not hard to see that if @xmath362}_{a , a^t}$ ] is odd@xmath223 , then so is @xmath362}_{a , b}$ ] where @xmath161 ( @xmath362}_{a , b}$ ] is just a subgraph of @xmath362}_{a , a^t}$ ] ) . on the other hand ,
if @xmath77 has more than one row , and there exist @xmath56 such that @xmath376 , then @xmath362}_{a , b}$ ] automatically fails condition c2 .
this follows because given any @xmath377 there exist two edges @xmath378 in @xmath362}_{a , b}$ ] , oppositely directed , of opposite sign , and one having label @xmath185 ; so @xmath362}_{a , b}$ ] has an e - cycle of length two which is not an s - cycle . in this paper
, most claims will involve matrix pairs @xmath243 satisfying @xmath379 ( namely , @xmath380 for some matrix @xmath215 ) , and often we will in fact assume @xmath161 .
[ example_rank1 ] if @xmath381 , we can easily compute @xmath362}_{a , a^t}$ ] to be : [ scale=.9 , place/.style = circle , draw = black!30,fill = black!0 , thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=<,shorten < = 2pt,>=stealth,semithick , post/.style = shorten > = 1pt,>=stealth,semithick ] ( 1t1 ) at ( 0,1 ) @xmath382 ; ( 1t2 ) at ( 2.5,0 ) @xmath383 ; ( 1t3 ) at ( 4,2 ) @xmath384 ; ( 1t4 ) at ( 2,4 ) @xmath385 ; ( 12 ) at ( 1.25 , 0.25 ) @xmath386 edge [ bend left = 5 , post ] ( 1t1 ) edge [ bend right = 5 , dashed , post ] ( 1t2 ) ; ( 12 ) at ( 1.9 , 1.8 ) @xmath387 edge [ bend right = 5 , post ] ( 1t1 ) edge [ bend left = 5 , dashed , post ] ( 1t3 ) ; ( 12 ) at ( 0.5 , 2.75 ) @xmath388 edge [ bend right = 8 , post ] ( 1t1 ) edge [ bend left = 8 , dashed , post ] ( 1t4 ) ; ( 12 ) at ( 3.5 , 0.75 ) @xmath389 edge [ bend left = 5 , post ] ( 1t2 ) edge [ bend right = 5 , dashed , post ] ( 1t3 ) ; ( 12 ) at ( 2.5 , 2.5 ) @xmath390 edge [ bend left = 5 , post , line width=0.08 cm , color = white ] ( 1t2 ) edge [ bend left = 5 , post ] ( 1t2 ) edge [ bend right = 5 , dashed , post ] ( 1t4 ) ; ( 12 ) at ( 3.25 , 3.25 ) @xmath391 edge [ bend left = 7 , post ] ( 1t3 ) edge [ bend right = 7 , dashed , post ] ( 1t4 ) ; all edge - labels are @xmath179 , so have been omitted .
more generally it is clear that if @xmath392 has no zero entries , then ( disregarding edge - signs and labels ) @xmath362}_{v , v^t}$ ] is a subdivision of the complete graph @xmath393 on the r - vertices @xmath394 .
results illustrating the use of the dsr@xmath44}$ ] graph [ sec : results1 ] here we develop and apply some results using dsr@xmath44}$ ] graphs before the development of further theory . some of the examples presented will later be shown to fall into families amenable to analysis , after the development of theory relating dsr and dsr@xmath44}$ ] graphs .
the following rather basic result about the second additive compounds of certain rank @xmath179 matrices can easily be proved directly , but demonstrates application of the dsr@xmath44}$ ] graph .
a far - reaching generalisation of this result will be presented as theorem [ thm : tree ] .
[ lem : columnvec ] if @xmath395 and @xmath396 , then @xmath397}$ ] is a @xmath89-matrix .
if the result is true for @xmath398 , then it follows for @xmath396 since the set of @xmath89-matrices is closed .
so consider @xmath398 . by remark [ remsignpat1 ] , @xmath362}_{s , v^t } = g^{[2]}_{s , s^t}$ ] .
note that from the definition of @xmath327 , row @xmath399 ( @xmath400 ) has at most two nonzero entries , the entry @xmath401 in column @xmath402 and the entry @xmath403 in column @xmath100 .
thus the edges incident with s - vertex @xmath404 have labels @xmath405 and @xmath406 . 1 .
no two cycles have odd intersection .
since no row of @xmath327 has more than two nonzero entries , each s - vertex in @xmath362}_{s , s^t}$ ] has maximum degree @xmath123 , as illustrated for example by the dsr@xmath44}$ ] graph of example [ example_rank1 ] .
this rules out odd intersections of cycles ( see remark [ remoddintersect ] ) .
@xmath362}_{s , s^t}$ ] is steady .
define @xmath407 , @xmath408 ( the set of all unordered pairs from @xmath409 ) .
let @xmath410 be distinct , and @xmath411 be distinct .
an arbitrary cycle in @xmath362}_{s , s^t}$ ] must be of the form @xmath412 where @xmath413 , and since @xmath414 are distinct this implies that @xmath415 .
( here we define @xmath416 . ) so the sequence of edge - labels in @xmath215 is @xmath417 and @xmath215 is clearly an @xmath331-cycle .
\(i ) and ( ii ) together imply that @xmath362}_{s , s^t}$ ] is odd@xmath223 , and hence by theorem [ mainp0thm ] that @xmath397}$ ] is a @xmath89-matrix .
[ ex3row ] we now return to the matrices in ( [ eq : motivate ] ) .
we will see that the dsr@xmath44}$ ] graph makes our earlier conclusions immediate , but shows moreover that the example provides just a special case of a general result , proposition [ 3sp1 ] below . using ( [ eq : matrep ] ) , we compute @xmath418 the dsr graphs @xmath228 ( left ) and @xmath362}_{a , a^t}$ ] ( right ) are depicted below : [ scale=0.75 , place/.style = circle , draw = blue!50,fill = blue!5 , thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.5 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , transition/.style = rectangle , draw = black!50,fill = black!5,thick , inner sep=0pt , minimum size=5.3 mm , pre/.style=<,shorten < = 2pt,>=stealth,semithick , post/.style = shorten > = 1pt,>=stealth,semithick ] ( 12 ) at ( -1.732,-1 ) @xmath190 ; ( 23 ) at ( 1.732,-1 ) @xmath192 ; ( 13 ) at ( 0 , 2 ) @xmath194 ; ( [ 13 ] ) at ( 1.732,1 ) @xmath419 edge [ bend right = 20 , post ] ( 13 ) edge [ bend left = 20 , post ] ( 23 ) ; ( [ 12 ] ) at ( 0,- 2 ) @xmath193 edge [ bend left = 20 , post ] ( 12 ) edge [ bend right = 20 , post ] ( 23 ) ; ( [ 11 ] ) at ( 0,0 ) @xmath191 edge [ post ] ( 12 ) edge [ post ] ( 13 ) edge [ post ] ( 23 ) ; [ scale=0.8 , place/.style = circle , draw = black!30,fill = black!0 , thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=<,shorten < = 2pt,>=stealth,semithick , post/.style = shorten > = 1pt,>=stealth,semithick ] ( 12 ) at ( -1.732,-1 ) @xmath386 ; ( 23 ) at ( 1.732,-1 ) @xmath389 ; ( 13 ) at ( 0 , 2 ) @xmath387 ; ( [ 13 ] ) at ( 1.732,1 ) @xmath384 edge [ dashed , bend right = 20 , post ] ( 13 ) edge [ dashed , bend left = 20 , post ] ( 23 ) ; ( [ 12 ] ) at ( 0,- 2 ) @xmath383 edge [ dashed , bend left = 20 , post ] ( 12 ) ; ( [ 11 ] ) at ( -1.732,1 ) @xmath382 edge [ bend right = 20 , post ] ( 12 ) ; ( [ 23 ] ) at ( 1.5 * 1.732,1.5 * 1 ) @xmath420 edge [ dashed , bend right = 30 , post ] ( 13 ) edge [ dashed , bend left = 30 , post ] ( 23 ) ; ( [ 22 ] ) at ( 1.5 * 0,- 1.5 * 2 ) @xmath421 edge [ dashed , bend left = 30 , post ] ( 12 ) edge [ bend right = 30 , post ] ( 23 ) ; ( [ 21 ] ) at ( -1.5 * 1.732,1.5 * 1 ) @xmath422 edge [ bend right = 30 , post ] ( 12 ) edge [ bend left = 30 , post ] ( 13 ) ; ( [ 33 ] ) at ( 2 * 1.732,2 * 1 ) @xmath423 edge [ dashed , bend left = 40 , post ] ( 23 ) ; ( [ 32 ] ) at ( 2 * 0,- 2 * 2 ) @xmath424 edge [ bend right = 40 , post ] ( 23 ) ; ( [ 31 ] ) at ( -2 * 1.732,2 * 1 ) @xmath425 edge [ bend right = 40 , post ] ( 12 ) edge [ bend left = 40 , post ] ( 13 ) ; all edge - labels are @xmath179 and have been omitted .
consequently , @xmath228 and @xmath362}_{a , a^t}$ ] are steady .
@xmath228 includes the e - cycles @xmath426 and @xmath427 which have odd intersection and so fails to be odd@xmath223 .
it is also easy to confirm that for some choice of @xmath202 , @xmath120 fails to be a @xmath89-matrix . on the other hand , since all r - vertices in @xmath362}_{a , a^t}$ ] have degree less than or equal to @xmath123 , @xmath362}_{a , a^t}$ ] can not have a pair of cycles with odd intersection , and so is odd@xmath223 ( see remark [ remoddintersect ] ) . by theorem
[ mainp0thm ] , @xmath284}$ ] is a @xmath89-matrix for all @xmath161 , and so , by lemma [ j2p0 ] , the nonreal eigenvalues of @xmath120 lie in @xmath86 .
[ rem : global_stability ] in fact , if @xmath202 , and the matrices @xmath428 arise as the jacobian matrices of a differential system , the form of @xmath165}$ ] in ( [ eq : j2_motivate ] ) has strong consequences . in particular ,
suppose @xmath429 is open , @xmath430 is @xmath10 , and consider the system @xmath431 on @xmath432 .
let @xmath433 be some compact , forward invariant , set , and assume that for each @xmath434 , @xmath435 is of the form in ( [ eq : motivate ] ) .
then , by observation of ( [ eq : j2_motivate ] ) , @xmath436}$ ] is irreducibly diagonally dominant ( in its columns ) with negative diagonal entries , and it is possible to find a ( single ) positive diagonal matrix @xmath437 such that @xmath438}\,e^{-1}$ ] is strictly diagonally dominant for all @xmath434 .
thus there exists a `` logarithmic norm '' @xmath439 and @xmath440 such that for all @xmath434 , @xmath441 } ) < p$ ] @xcite , and applying the theory developed by li and muldowney ( @xcite for example ) , the system @xmath431 can not have invariant closed curves in @xmath442 : periodic orbits whether stable or unstable are ruled out , as are homoclinic trajectories , heteroclinic cycles , etc . in fact every nonwandering point of the system must be an equilibrium @xcite .
example [ example_motivate ] suggests the following much more general result : [ 3sp1 ] ( i ) consider any @xmath443 @xmath444 matrix @xmath77 and any @xmath161 .
then @xmath284}$ ] is a @xmath89-matrix and the nonreal spectrum of @xmath120 does not intersect @xmath83 .
( ii ) consider any @xmath445 @xmath444 matrix @xmath77 and any @xmath161 .
then the nonreal spectrum of @xmath120 does not intersect @xmath83 .
if the results hold for each @xmath202 , they follow for each @xmath161 by closure .
choose any @xmath202 .
( i ) all edge - labels in @xmath362}_{a , b}$ ] are @xmath179 and so @xmath362}_{a , b}$ ] is steady .
on the other hand , all r - vertices in @xmath362}_{a , b}$ ] have degree less than or equal to two ( three edges of the form @xmath446 , @xmath447 and @xmath448 can not all exist ) , and so odd intersections between cycles are impossible . consequently , @xmath362}_{a , b}$ ] is odd@xmath223 .
the conclusions follow from theorem [ mainp0thm ] and lemma [ j2p0 ] .
( ii ) from ( i ) , @xmath449}$ ] is a @xmath89-matrix and consequently , by lemma [ j2p0 ] , the nonreal spectrum of @xmath450 does not intersect @xmath83 . since the nonzero spectrum of @xmath120 is equal to that of @xmath450 ( lemma [ spectranposelemma ] ) , the same holds for @xmath120 .
further generalisations of proposition [ 3sp1 ] will be proved as theorem [ thm:3sp ] later .
[ rem : crn3nohopf ] proposition [ 3sp1 ] has the following interpretation in terms of the dynamics of crns . consider a system of reactions where all entries in the stoichiometric matrix have magnitude @xmath179 ( a fairly common situation ) and where the kinetics satisfies the condition in ( [ eq : crnjac ] ) .
suppose further there are no more than three substrates , or no more than three reactions .
then the system is incapable of hopf bifurcations .
the next result which follows with little effort , involves @xmath444 matrices with no more than two entries in each column ( or row ) .
[ prop : switchrs ] let @xmath77 be an @xmath35 @xmath444 matrix .
( i ) if @xmath77 has no more than two nonzero entries in each column then , for each @xmath161 , @xmath120 and @xmath284}$ ] are @xmath89-matrices and so @xmath120 is positive semistable .
( ii ) if @xmath77 has no more than two nonzero entries in each row then , for any @xmath161 , @xmath120 is positive semistable .
if the results are true for each @xmath202 , then they hold for each @xmath161 by closure .
choose any @xmath202 .
( i ) as all edge - labels in @xmath186 and @xmath362}_{a , b}$ ] are @xmath179 , both are steady .
as all r - vertices in @xmath186 have degree less than or equal to two , by remark [ remoddintersect ] , @xmath186 is in fact odd@xmath223 .
further , all r - vertices in @xmath362}_{a , b}$ ] have degree less than or equal to two : this follows by noting that the edge @xmath451 exists if and only if @xmath452 , which ( for fixed @xmath100 , @xmath402 ) can occur for a maximum of two indices @xmath26 by the assumption that no column of @xmath77 contains more than two nonzero entries
. consequently , @xmath362}_{a , b}$ ] is odd@xmath223 .
the remaining conclusions follow from theorem [ mainp0thm ] and lemma [ lemposstable0 ] .
( ii ) from ( i ) , @xmath450 and @xmath449}$ ] are @xmath89-matrices and consequently , by lemma [ lemposstable0 ] , @xmath450 is positive semistable .
since the nonzero spectrum of @xmath120 is equal to that of @xmath450 ( lemma [ spectranposelemma ] ) , the same holds for @xmath120 .
proposition [ prop : switchrs ] will be generalised as theorem [ thm : alls ] , where the requirement that @xmath77 is a @xmath444 matrix will be weakened to the requirement that @xmath228 is steady .
crns where the degree of each s - vertex is less than or equal to two were discussed in @xcite .
returning to this example we have the matrix @xmath77 and dsr graph @xmath228 : [ scale=0.95 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.5 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.5 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; at ( -4.5,0 ) @xmath453 ; ( r1 ) at ( 180:1 ) @xmath193 ; ( s2 ) at ( 90:1 ) @xmath192 edge[pre1,dashed , bend right=25 , - ] node [ auto , swap]@xmath454 ( r1 ) ; ( r2 ) at ( 0:1 ) @xmath191 edge[pre1,bend right=25 , - ] node [ auto , swap]@xmath455(s2 ) ; ( s1 ) at ( -90:1 ) @xmath190
edge[pre1,bend right=25 , - ] node [ auto , swap]@xmath456(r2 ) edge[pre1,bend left=25 , - ] node [ auto]@xmath457(r1 ) ; ( s3 ) at ( -4.7,0 ) @xmath194 edge[pre1,- ] node [ auto , swap]@xmath458(r2 ) ; ( r3 ) at ( -3.3,0 ) @xmath419 edge[pre1,- ] node [ auto , swap]@xmath459(s3 ) ; ( s4 ) at ( -1.9,0 ) @xmath460 edge[pre1,- ] node [ auto , swap]@xmath461(r3 ) ; @xmath228 is odd by inspection , and so @xmath120 is a @xmath89-matrix for all @xmath161 . following removal of vertices and edges which do not participate in any cycles , we get the reduced representation of @xmath362}_{a , a^t}$ ] : [ scale=1.2 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; ( 2t3 ) at ( 180:1 ) @xmath420 ; ( 13 ) at ( 90:1 ) @xmath387 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath456 ( 2t3 ) ; ( 1t3 ) at ( 0:1 ) @xmath384 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath457 ( 13 ) ; ( 23 ) at ( -90:1 ) @xmath389
edge[pre1,bend right=25 , - ] node [ fill = white]@xmath454 ( 1t3 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath455 ( 2t3 ) ; ( 2t4 ) at ( 180:1 ) @xmath462 ; ( 14 ) at ( 90:1 ) @xmath388 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath456 ( 2t4 ) ; ( 1t4 ) at ( 0:1 ) @xmath385 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath457 ( 14 ) ; ( 24 ) at ( -90:1 ) @xmath390
edge[pre1,bend right=25 , - ] node [ fill = white]@xmath454 ( 1t4 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath455 ( 2t4 ) ; ( 3t1 ) at ( -2,1.2 ) @xmath425
edge[pre1,bend right=5 , - ] node [ fill = white]@xmath459 ( 13 ) edge[pre1,bend left=5 , - ] node [ fill = white]@xmath461 ( 14 ) ; ( 3t2 ) at ( -2,-1.2 ) @xmath424
edge[pre1,bend left=5 , - ] node [ fill = white]@xmath459 ( 23 ) edge[pre1,bend right=5 , - ] node [ fill = white]@xmath461 ( 24 ) ; \(12 ) at ( -7,0 ) @xmath386 ; ( 2t1 ) at ( -6,1.2 ) @xmath422 edge[pre1,bend left=15 , - ] node [ fill = white]@xmath458 ( 13 ) edge[pre1,bend right=25 , - ] node [ fill = white]@xmath455 ( 12 ) ; ( 2t2 ) at ( -6,-1.2 ) @xmath421
edge[pre1,bend right=15 , - ] node [ fill = white]@xmath458 ( 23 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath456 ( 12 ) ; which can be computed to be odd@xmath223 .
consequently @xmath284}$ ] is a @xmath89-matrix for all @xmath161 .
thus , by lemma [ lemposstable0 ] , @xmath120 is positive stable for each @xmath161 .
[ example_motivate1 ] we illustrate the use of the dsr@xmath44}$ ] graph to make claims about the second additive compound of a single square matrix ( rather than a product of two matrices ) .
let @xmath463 and consider the matrix and dsr graph : [ scale=1.5 , place/.style = circle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5.5 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten < = -1pt , shorten > = 0pt,>=stealth,semithick ] ; at ( -4.5,0 ) @xmath464 ; at ( -1.8,0 ) @xmath465 ; ( r2 ) at ( -135:1 ) @xmath191 ; ( s2 ) at ( -90:1 ) @xmath192 edge[pre , bend left=15 , - ] node [ auto ] ( r2 ) ; ( r3 ) at ( -45:1 ) @xmath419 edge[pre , bend left=15 , - > ] node [ auto](s2 ) ; ( s3 ) at ( 0:1 ) @xmath194 edge[pre , bend left=15 , - ] node [ auto](r3 ) ; ( r4 ) at ( 45:1 ) @xmath466 edge[pre , bend left=15 , - > ] node [ auto](s3 ) ; ( s4 ) at ( 90:1 ) @xmath460 edge[pre , bend left=15 , - ] node [ auto](r4 ) ; ( r1 ) at ( 135:1 ) @xmath193 edge[pre , bend left=15 , - > ] node [ auto](s4 ) edge[pre , dashed , bend left=20 , - > ] node [ auto](s2 ) ; ( s1 ) at ( 180:1 ) @xmath190 edge[pre , bend left=15 , - ] node [ auto](r1 ) edge[pre1,bend right=15 , < - ] node [ auto , swap](r2 ) ; here @xmath467 is the @xmath468 identity matrix which we note lies in @xmath469 .
@xmath470 has exactly two cycles , one of which is odd and one of which is even . following removal of vertices and edges which do not participate in any cycles , we get the reduced representation of @xmath362}_{a , i}$ ] : [ scale=1.1 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; ( 2t3 ) at ( 180:1 ) @xmath420 ; ( 13 ) at ( 90:1 ) @xmath387 edge[pre1,dashed , bend right=25 , < - ] ( 2t3 ) ; ( 1t3 ) at ( 0:1 ) @xmath384 edge[pre1,dashed , bend right=25 , - ] ( 13 ) ; ( 23 ) at ( -90:1 ) @xmath389
edge[pre1,bend right=25 , < - ] ( 1t3 ) edge[pre1,dashed , bend left=25 , - ] ( 2t3 ) ; ( 2t4 ) at ( 90:1 ) @xmath462 ; ( 14 ) at ( 0:1 ) @xmath388 edge[pre1,dashed , bend right=25 , < - ] ( 2t4 ) ; ( 1t4 ) at ( -90:1 ) @xmath385 edge[pre1,dashed , bend right=25 , - ] ( 14 ) ; ( 24 ) at ( 180:1 ) @xmath390 edge[pre1,bend right=25 , < - ] ( 1t4 ) edge[pre1,dashed , bend left=25 , - ] ( 2t4 ) ; \(34 ) at ( -4.4,0 ) @xmath391 edge[pre1,<- ] ( 1t3 ) ; ( 3t4 ) at ( -2.7,0 ) @xmath471 edge[pre1,dashed,- ] ( 34 ) edge[pre1,dashed,- > ] ( 24 ) ; ( 3t1 ) at ( -5.3,1.05 ) @xmath425
edge[pre1,bend right=3,- ] ( 13 ) ; ( 12 ) at ( -3.8,1 ) @xmath386 edge[pre1,bend right=3,<- ] ( 3t1 ) ; ( 1t2 ) at ( -2.3,0.8 ) @xmath383 edge[pre1,dashed , bend right=3,- ] ( 12 ) edge[pre1,bend left=10,- > ] ( 24 ) ; ( 4t2 ) at ( -3.8,-1 ) @xmath472
edge[pre1,bend left=3,- > ] ( 23 ) edge[pre1,bend right=15,- ] ( 24 ) ; ( 4t1 ) at ( -1.5,1.5 ) @xmath473 ; ( 4t1 ) .. controls ( -6,1.5 ) and ( -6.5 , 1.5 ) .. ( 13 ) ; ( 4t1 ) .. controls ( 0,1.5 ) and ( 0.9 , 1.4 ) .. ( 14 ) ; edge labels have been omitted for the reason that @xmath362}_{a , i}$ ] is odd , namely there are no e - cycles at all .
consequently @xmath474 } = ( ai)^{[2]}$ ] is a @xmath89-matrix , and @xmath77 can not have a pair of nonreal eigenvalues in @xmath83 .
[ rem : monotonecycles ] the matrix @xmath77 in example [ example_motivate1 ] is not obviously amenable to analysis using the results of @xcite or @xcite : the sufficient conditions for the absence of _ stable _ oscillation in dynamical systems developed in these papers are neither necessary nor sufficient for the preclusion of hopf bifurcation .
relationships between the dsr and dsr@xmath44}$ ] graphs [ sec : theory ] throughout this section we consider two matrices @xmath475 and @xmath168 ( @xmath359 ) . in order to understand better the dsr@xmath44}$ ] graph , it is convenient to develop some theory on the relationship between structures in @xmath169 and @xmath362}_{a , b}$ ] .
the results presented below begin this process .
applications of this theory are presented in the final section . removing leaves
our first observation is that a certain simple operation on dsr graphs of the form @xmath228 the removal of pendant r - vertices and their incident edges does not alter the cycle structure of the corresponding dsr@xmath44}$ ] graph .
this claim does not extend to the removal of pendant s - vertices .
[ lem : leafremove ] consider an @xmath35 matrix @xmath77 ( @xmath476 ) whose @xmath100th column has a single nonzero entry @xmath477 , so that vertex @xmath478 in @xmath228 has degree @xmath179 .
let @xmath285 be the submatrix of @xmath77 where the @xmath100th column has been removed . then @xmath362}_{\tilde a , \tilde a^t}$ ] is odd@xmath223 ( resp .
odd , resp .
steady ) if and only if @xmath362}_{a , a^t}$ ] is odd@xmath223 ( resp .
odd , resp .
steady ) . in one direction
the result is trivial : if @xmath362}_{a , a^t}$ ] is odd@xmath223 ( resp .
odd , resp .
steady ) , then so is @xmath362}_{\tilde a , \tilde a^t}$ ] as it is a subgraph of @xmath362}_{a , a^t}$ ] . in the other direction , we show that @xmath362}_{\tilde a , \tilde a^t}$ ] is an induced subgraph of @xmath362}_{a , a^t}$ ] obtained by removing only leaves of @xmath362}_{a , a^t}$ ] , and consequently has the same cycle structure . note that @xmath362}_{\tilde a , \tilde a^t}$ ] is the induced subgraph of @xmath362}_{a , a^t}$ ] obtained by removing all r - vertices of the form @xmath479 and their incident edges .
however , an edge in @xmath362}_{a , a^t}$ ] of the form @xmath451 exists if and only if @xmath480 which occurs iff @xmath481 .
thus for each @xmath402 , column @xmath479 of @xmath301 has a single nonzero entry ( in row @xmath482 ) , and the corresponding r - vertex in @xmath362}_{a , a^t}$ ] is a leaf which can be removed without affecting any cycles of @xmath362}_{a , a^t}$ ] .
projections of dsr@xmath44}$ ] cycles it is useful to distinguish two types of edges in a dsr@xmath44}$ ] graph , @xmath483 and @xmath484 . recall from section [ sec : c2decomp ] that if @xmath324 then @xmath485 this fact , in conjunction with the definition of @xmath169 , implies that for any edge @xmath486 in @xmath362}_{a , b}$ ] there exists an edge @xmath487 in @xmath169 having the same direction , the same label and the same sign ; and similarly , for any edge @xmath451 in @xmath362}_{a , b}$ ] there exists an edge @xmath488 in @xmath186 having the same direction and label , but _
opposite sign_. we have defined a mapping of the edge - set of @xmath362}_{a , b}$ ] into the edge - set of @xmath169 . [
def : proj ] let @xmath489 and @xmath490}_{a , b})$ ] denote the set of edges of @xmath169 and @xmath362}_{a , b}$ ] respectively . similarly let @xmath491 and @xmath492}_{a , b})$ ] .
the mapping @xmath493 defined by @xmath494 is termed the _ projection map _ on edges of _ dsr_@xmath44}.$ ] [ rem : pi ] a few immediate properties of the projection map are worth emphasising .
@xmath495 is surjective if @xmath496 , but not necessarily injective if @xmath497 . indeed , if @xmath498 then for any @xmath499 we have @xmath500 .
if there is more than one choice for such @xmath402 then @xmath495 is not injective .
\2 . following the discussion preceding definition [ def : proj ] , @xmath495 acts on edge - signs as follows : @xmath501 \3 .
@xmath495 does not extend naturally to vertices of @xmath362}_{a , b}$ ] .
while @xmath495 may be viewed as mapping the r - vertex @xmath502 to @xmath503 , this interpretation is ambiguous for @xmath504-vertices @xmath505 : edges in @xmath506 incident to @xmath369 may be mapped to edges in @xmath437 incident to either @xmath507 or to @xmath508 . as a consequence , a walk in @xmath362}_{a , b}$ ] does not necessarily project to a walk in @xmath169 .
this is illustrated in the following diagram , which also shows that projection may change edge signs .
[ scale=2.6 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.5 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , post/.style = shorten > = 1pt , shorten > = 1pt,>=stealth,semithick ] ; ( 13 ) at ( -2,0 ) @xmath387 ; ( 12 ) at ( 0,0 ) @xmath386 ; ( [ 11 ] ) at ( -1,0 ) @xmath425 edge[pre1,- ] ( 13 ) edge [ pre ] ( 12 ) ; ( [ 21 ] ) at ( 1,0 ) @xmath421 edge [ < - , dashed , post ] ( 12 ) ; \(13 ) at ( -2,-.7 ) @xmath194 ; ( 12 ) at ( 0,-.7 ) @xmath192 ; ( [ 11 ] ) at ( -1,-.7 ) @xmath419 edge[pre1,- ] ( 13 ) edge [ pre ] ( 12 ) ; ( -1.5,-.1 ) ( -1.5,-.6 ) ; at ( -1.6 , -.3 ) @xmath495 ; ( -.5,-.1 ) ( -.5,-.6 ) ; at ( -.6 , -.3 ) @xmath495 ; ( .5,-.1 ) ( 1,-.6 ) ; at ( .6 , -.3 ) @xmath495 ; \(14 ) at ( .5,-.7 ) @xmath190 ; ( [ 14 ] ) at ( 1.5,-.7 ) @xmath191 edge[->,pre1 ] ( 14 ) ; having defined the projection map , it is natural to examine the projection of cycles in @xmath362}_{a , b}$ ] .
[ prop : proj ] let @xmath509 be a cycle in @xmath362}_{a , b}$ ] with edges enumerated so that the initial vertex of @xmath510 is an s - vertex @xmath369 .
then there exists a partition of @xmath511 such that @xmath512 and @xmath513 satisfy either : \(i ) @xmath514 and @xmath515 are closed walks , one containing @xmath507 and the other containing @xmath516 or \(ii ) @xmath514 and @xmath515 are walks , one from @xmath507 to @xmath508 and the other from @xmath508 to @xmath507 .
let @xmath517 denote the s - vertices of @xmath215 , and such that @xmath518 for each @xmath26 .
the construction of @xmath514 and @xmath515 is done by successively appending the walks of length two @xmath519 of @xmath186 to either @xmath514 or @xmath515 .
begin by defining @xmath520 and @xmath521 as the empty one - vertex walks @xmath522 and @xmath523 , respectively .
next , inductively define , for @xmath524 , walks @xmath525 and @xmath526 in @xmath186 whose end vertices are different elements of @xmath527 in the following fashion .
assume that @xmath528 is the end vertex of @xmath529 and @xmath530 is the end vertex of @xmath531 ( the other case is similar ) .
the walk @xmath532 has the form @xmath533 , where one of the following possibilities holds : ( 1 ) @xmath534 so that @xmath535 ; ( 2 ) @xmath536 so that @xmath537 ; ( 3 ) @xmath538 , so that @xmath539 ; ( 4 ) @xmath540 so that @xmath541 . in cases ( 1 ) and
( 2 ) define @xmath542 and let @xmath543 ; in cases ( 3 ) and ( 4 ) define @xmath544 and let @xmath545 . either way , @xmath546 and @xmath547 are walks ending at different vertices from @xmath527 .
it is easy to check that @xmath548 and @xmath549 satisfy the conclusion of the proposition .
[ rem : ww ] 1 . in case ( i ) of proposition [ prop : proj ] @xmath514 and @xmath515 are closed walks in @xmath169 and it is not hard to see that choosing a different initial vertex from @xmath215 in @xmath362}_{a , b}$ ] leads to the same two closed walks ( albeit possibly renamed , and traversed from different initial points ) ; in case ( ii ) of proposition [ prop : proj ] @xmath514 and @xmath515 traversed successively form a closed walk of length @xmath550 , which we will denote by @xmath551 again choosing a different initial vertex from @xmath215 leads to the same object , traversed from a different initial point . \2 . by construction ,
a pair of consecutive edges of @xmath215 beginning and terminating at an s - vertex are projected to a pair of consecutive edges in either @xmath514 or @xmath515 .
with notation from proposition [ prop : proj ] , this means that @xmath552 for each @xmath553 , and @xmath554 for each @xmath555 .
in particular , a sequence @xmath556 traversing an unoriented edge @xmath557 in each direction consecutively can not occur in @xmath558 or @xmath559 : such a sequence must be the projection of a sequence of edges in @xmath362}_{a , b}$ ] of the form @xmath560 , traversing an unoriented edge @xmath561 back and forth , which clearly can not occur in a cycle @xmath215 .
it is transparent from the previous remark that whether projection of a cycle @xmath215 in @xmath362}_{a , b}$ ] leads to case ( i ) or case ( ii ) in proposition [ prop : proj ] is a property of @xmath215 alone . a cycle @xmath215 of @xmath362}_{a , b}$ ]
will be called * direct * if it projects as in case ( i ) and * twisted * if it projects as in case ( ii ) . in view of remark [ rem : ww ] the following notion is well - defined : [ def : projcycle ] let @xmath215 be a cycle in @xmath362}_{a , b}$ ] and @xmath562 corresponding walks in @xmath169 constructed as in proposition [ prop : proj ] .
the _ projection _ @xmath563 of @xmath215 is defined as follows : \(i ) if @xmath215 is direct then @xmath564 \(ii ) if @xmath215 is twisted then @xmath565 .
[ example : directtwisted ] we return to the dsr@xmath44}$ ] and dsr graphs in example [ example_motivate0 ] to illustrate the projections of direct and twisted cycles . the cycle @xmath566 ( with image shaded in the dsr@xmath44}$ ] graph on the left ) is direct , and projects to the two closed walks @xmath567 ( with images shaded in the dsr graph on the right ) .
[ scale=0.93 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; ( 2t3 ) at ( 180:1 ) @xmath420 ; ( 13 ) at ( 90:1 ) @xmath387 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath456 ( 2t3 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( 2t3 ) ; ( 1t3 ) at ( 0:1 ) @xmath384 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath457 ( 13 ) ; ( 23 ) at ( -90:1 ) @xmath389
edge[pre1,bend right=25 , - ] node [ fill = white]@xmath454 ( 1t3 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath455 ( 2t3 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend left=25 , - ] ( 2t3 ) ; ( 2t4 ) at ( 180:1 ) @xmath462 ; ( 14 ) at ( 90:1 ) @xmath388 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath456 ( 2t4 ) ; ( 1t4 ) at ( 0:1 ) @xmath385 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath457 ( 14 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( 14 ) ; ( 24 ) at ( -90:1 ) @xmath390
edge[pre1,bend right=25 , - ] node [ fill = white]@xmath454 ( 1t4 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( 1t4 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath455 ( 2t4 ) ; ( 3t1 ) at ( -2,1.2 ) @xmath425
edge[pre1,bend right=5 , - ] node [ fill = white]@xmath459 ( 13 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=5 , - ] ( 13 ) edge[pre1,bend left=5 , - ] node [ fill = white]@xmath461 ( 14 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend left=5 , - ] ( 14 ) ; ( 3t2 ) at ( -2,-1.2 ) @xmath424
edge[pre1,bend left=5 , - ] node [ fill = white]@xmath459 ( 23 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend left=5 , - ] ( 23 ) edge[pre1,bend right=5 , - ] node [ fill = white]@xmath461 ( 24 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=5 , - ] ( 24 ) ; \(12 ) at ( -7,0 ) @xmath386 ; ( 2t1 ) at ( -6,1.2 ) @xmath422
edge[pre1,bend left=15 , - ] node [ fill = white]@xmath458 ( 13 ) edge[pre1,bend right=25 , - ] node [ fill = white]@xmath455 ( 12 ) ; ( 2t2 ) at ( -6,-1.2 ) @xmath421
edge[pre1,bend right=15 , - ] node [ fill = white]@xmath458 ( 23 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath456 ( 12 ) ; [ scale=0.75 , place/.style = circle , draw = black!7,fill = black!0,thick , inner sep=0pt , minimum size=5.5 mm , transition/.style = rectangle , draw = black!7,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.5 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; ( r1 ) at ( 180:1 ) @xmath193 ; ( s2 ) at ( 90:1 ) @xmath192 edge[pre1,dashed , bend right=25 , - ] node [ auto , swap]@xmath454 ( r1 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( r1 ) ; ( r2 ) at ( 0:1 ) @xmath191 edge[pre1,bend right=25 , - ] node [ auto , swap]@xmath455(s2 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( s2 ) ; ( s1 ) at ( -90:1 ) @xmath190
edge[pre1,bend right=25 , - ] node [ auto , swap]@xmath456(r2 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( r2 ) edge[pre1,bend left=25 , - ] node [ auto]@xmath457(r1 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend left=25 , - ] ( r1 ) ; ( s3 ) at ( -4.7,0 ) @xmath194 edge[pre1,- ] node [ auto , swap]@xmath458(r2 ) ; ( r3 ) at ( -3.3,0 ) @xmath419 edge[pre1,- ] node [ auto , swap]@xmath459(s3 ) edge[color = black!40,opacity=0.5,line width=0.2cm,- ] ( s3 ) ; ( s4 ) at ( -1.9,0 ) @xmath460 edge[pre1,- ] node [ auto , swap]@xmath461(r3 ) edge[color = black!40,opacity=0.5,line width=0.2cm,- ] ( r3 ) ; on the other hand , the cycle @xmath568 ( with image shaded in the dsr@xmath44}$ ] graph on the left ) is twisted , and projects to the closed walk @xmath569 ( with image shaded in the dsr graph on the right ) .
[ scale=0.93 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; ( 2t3 ) at ( 180:1 ) @xmath420 ; ( 13 ) at ( 90:1 ) @xmath387 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath456 ( 2t3 ) ; ( 1t3 ) at ( 0:1 ) @xmath384 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath457 ( 13 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( 13 ) ; ( 23 ) at ( -90:1 ) @xmath389
edge[pre1,bend right=25 , - ] node [ fill = white]@xmath454 ( 1t3 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( 1t3 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath455 ( 2t3 ) ; ( 2t4 ) at ( 180:1 ) @xmath462 ; ( 14 ) at ( 90:1 ) @xmath388 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath456 ( 2t4 ) ; ( 1t4 ) at ( 0:1 ) @xmath385 edge[pre1,dashed , bend right=25 , - ] node [ fill = white]@xmath457 ( 14 ) ; ( 24 ) at ( -90:1 ) @xmath390
edge[pre1,bend right=25 , - ] node [ fill = white]@xmath454 ( 1t4 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath455 ( 2t4 ) ; ( 3t1 ) at ( -2,1.2 ) @xmath425
edge[pre1,bend right=5 , - ] node [ fill = white]@xmath459 ( 13 ) edge[pre1,bend left=5 , - ] node [ fill = white]@xmath461 ( 14 ) ; ( 3t2 ) at ( -2,-1.2 ) @xmath424
edge[pre1,bend left=5 , - ] node [ fill = white]@xmath459 ( 23 ) edge[pre1,bend right=5 , - ] node [ fill = white]@xmath461 ( 24 ) ; \(12 ) at ( -7,0 ) @xmath386 ; ( 2t1 ) at ( -6,1.2 ) @xmath422 edge[pre1,bend left=15 , - ] node [ fill = white]@xmath458 ( 13 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend left=15 , - ] ( 13 ) edge[pre1,bend right=25 , - ] node [ fill = white]@xmath455 ( 12 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( 12 ) ; ( 2t2 ) at ( -6,-1.2 ) @xmath421
edge[pre1,bend right=15 , - ] node [ fill = white]@xmath458 ( 23 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=15 , - ] ( 23 ) edge[pre1,dashed , bend left=25 , - ] node [ fill = white]@xmath456 ( 12 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend left=25 , - ] ( 12 ) ; [ scale=0.75 , place/.style = circle , draw = blue!50,fill = blue!5,thick , inner sep=0pt , minimum size=5.5 mm , transition/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.5 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; ( r1 ) at ( 180:1 ) @xmath193 ; ( s2 ) at ( 90:1 ) @xmath192 edge[pre1,dashed , bend right=25 , - ] node [ auto , swap]@xmath454 ( r1 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( r1 ) ; ( r2 ) at ( 0:1 ) @xmath191 edge[pre1,bend right=25 , - ] node [ auto , swap]@xmath455(s2 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( s2 ) ; ( s1 ) at ( -90:1 ) @xmath190
edge[pre1,bend right=25 , - ] node [ auto , swap]@xmath456(r2 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend right=25 , - ] ( r2 ) edge[pre1,bend left=25 , - ] node [ auto]@xmath457(r1 ) edge[color = black!40,opacity=0.5,line width=0.2cm , bend left=25 , - ] ( r1 ) ; ( s3 ) at ( -4.7,0 ) @xmath194 edge[pre1,- ] node [ auto , swap]@xmath458(r2 ) edge[color = black!40,opacity=0.5,line width=0.2cm,- ] ( r2 ) ; ( r3 ) at ( -3.3,0 ) @xmath419 edge[pre1,- ] node [ auto , swap]@xmath459(s3 ) ; ( s4 ) at ( -1.9,0 ) @xmath460 edge[pre1,- ] node [ auto , swap]@xmath461(r3 ) ; [ rem : extendpi ] let @xmath215 be a cycle in @xmath362}_{a , b}$ ]
if @xmath215 is twisted then @xmath570 whereas if @xmath215 is direct and @xmath571 then @xmath572
. \2 . in the case
that @xmath215 is direct , various scenarios are possible : one of @xmath514 or @xmath515 may be empty ; @xmath514 and @xmath515 may be disjoint ; they may share vertices or edges ; or they might even coincide . to see how one of @xmath514 or @xmath515 may be empty ,
consider the following example .
let @xmath573 be such that @xmath574 are positive .
we get an 8-cycle @xmath575 in @xmath362}_{a , a^t}$ ] which projects to an empty walk @xmath514 and an 8-cycle @xmath576 in @xmath228 : [ scale=1.2 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.3 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten < = -1pt , shorten > = 0pt,>=stealth,semithick ] ; at ( -2,0 ) @xmath577 ; ( 12 ) at ( -1,1 ) @xmath386 ; ( 23 ) at ( 1,1 ) @xmath389 ; ( 25 ) at ( -1,-1 ) @xmath578 ; ( 24 ) at ( 1,-1 ) @xmath390 ; ( [ 12 ] ) at ( 0,1 ) @xmath383 edge[pre,- , dashed ] node [ auto , swap]@xmath579 ( 12 ) edge [ pre,- ] node [ auto]@xmath580 ( 23 ) ; ( [ 22 ] ) at ( 1,0 ) @xmath421 edge[pre,- ] node [ auto , swap]@xmath581 ( 23 ) edge [ pre,- ] node [ auto]@xmath582 ( 24 ) ; ( [ 32 ] ) at ( 0,-1 ) @xmath424 edge[pre,- ] node [ auto , swap]@xmath583 ( 24 ) edge [ pre,- ] node [ auto]@xmath584 ( 25 ) ; ( [ 42 ] ) at ( -1,0 ) @xmath472 edge[pre,- ] node [ auto , swap]@xmath585 ( 25 ) edge [ pre,- , dashed ] node [ auto]@xmath586 ( 12 ) ; at ( 3,0 ) @xmath587 ; at ( -2,0 ) @xmath588 ; ( 12 ) at ( -1,1 ) @xmath190 ; ( 23 ) at ( 1,1 ) @xmath194 ; ( 25 ) at ( -1,-1 ) @xmath589 ; ( 24 ) at ( 1,-1 ) @xmath460 ; ( [ 12 ] ) at ( 0,1 ) @xmath193 edge[pres,- ] node [ auto , swap]@xmath579 ( 12 ) edge [ pres,- ] node [ auto]@xmath580 ( 23 ) ; ( [ 22 ] ) at ( 1,0 ) @xmath191 edge[pres,- ] node [ auto , swap]@xmath581 ( 23 ) edge [ pres,- ] node [ auto]@xmath582 ( 24 ) ; ( [ 32 ] ) at ( 0,-1 ) @xmath419 edge[pres,- ] node [ auto , swap]@xmath583 ( 24 ) edge [ pres,- ] node [ auto]@xmath584 ( 25 ) ; ( [ 42 ] ) at ( -1,0 ) @xmath466 edge[pres,- ] node [ auto , swap]@xmath585 ( 25 ) edge [ pres,- ] node [ auto]@xmath586 ( 12 ) ; note that in this case , the index 2 is common to all s - vertices in @xmath215 , and , disregarding signs of edges , the edge - sequence of @xmath515 mirrors that of @xmath215 . in situations such as this , we make an abuse of terminology and write @xmath590 rather than @xmath591 to see how @xmath514 and @xmath515 may coincide consider the 20-cycle @xmath215 illustrated below . for simplicity only s - vertices
are depicted : [ scale=1.6 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!50,fill = black!5,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.3 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; at ( -1.5,0 ) @xmath577 ; ( 0,0 ) circle ( 1 cm ) ; ( 12 ) at ( 180:1 ) @xmath386 ; ( 13 ) at ( 144:1 ) @xmath387 ; ( 23 ) at ( 108:1 ) @xmath389 ; ( 24 ) at ( 72:1 ) @xmath390 ; ( 34 ) at ( 36:1 ) @xmath391 ; ( 35 ) at ( 0:1 ) @xmath592 ; ( 45 ) at ( -36:1 ) @xmath593 ; ( 14 ) at ( -72:1 ) @xmath388 ; ( 15 ) at ( -108:1 ) @xmath594 ; ( 25 ) at ( -144:1 ) @xmath578 ; at ( -2,0 ) @xmath595 ; ( 0,0 ) circle ( 1 cm ) ; ( 12 ) at ( 180:1 ) @xmath190 ; ( 13 ) at ( 106:1 ) @xmath192 ; ( 23 ) at ( 36:1 ) @xmath194 ; ( 24 ) at ( -36:1 ) @xmath460 ; ( 34 ) at ( -106:1 ) @xmath589 ; * identifying whether a cycle is direct or twisted . *
if @xmath369 and @xmath596 are two consecutive s - vertices of a cycle @xmath215 in @xmath362}_{a , b}$ ] , then @xmath597 must include at least one element , say @xmath26 .
we call the pair @xmath598 an * inversion * if @xmath599 . whether a cycle is direct or twisted can be computed by counting inversions : [ prop : parityinv ] a cycle @xmath215 of a dsr@xmath44}$ ] graph @xmath362}_{a , b}$ ] is direct if it has an even number of inversions , and twisted otherwise .
let @xmath215 be a direct cycle with the sequence of s - vertices @xmath600 .
assume that @xmath601 for all @xmath100 . by the construction in proposition [ prop : proj ] , one element of @xmath602
is assigned to @xmath514 and the other is assigned to @xmath515 .
define @xmath603 as @xmath604 note that @xmath605 if and only if the pair @xmath606 is an inversion . by definition , @xmath215 is direct if and only if @xmath514 is a closed walk , i.e. , if @xmath607 which is equivalent to saying that the number of switches between 1 and 2 in the sequence @xmath608 is even .
the nature of the two cycles in example [ example : directtwisted ] is now easily computed : the first contains no inversions and so is direct ; the second contains one inversion @xmath609 and is therefore twisted .
the parity of a cycle in @xmath362}_{a , b}$ ] relates to the parity of its projection , as shown in the following result : [ prop : parity ] let @xmath215 denote a cycle in @xmath362}_{a , b}$ ] .
\(i ) if @xmath215 is direct and @xmath571 then @xmath610 \(ii ) if @xmath215 is twisted then @xmath611 let @xmath509 be such that the initial vertex of @xmath510 is an s - vertex , and let @xmath612 , @xmath613 be the partition of @xmath614 obtained as in proposition [ prop : proj ] . in view of remark [ rem : ww ] we have @xmath615.\ ] ] ( we adopt the convention that an empty product has value @xmath179 . ) equation ( [ eq : projsign ] ) implies that , for any @xmath616 @xmath617 if the two s - vertices of the pair @xmath618 form an inversion , and @xmath619 otherwise . denoting by @xmath620 the number of pairs from @xmath621 whose
s - vertices form an inversion , equation ( [ eq : paritypf ] ) becomes @xmath622 multiplying with the corresponding equation for @xmath515 yields @xmath623 where @xmath624 is the number of inversions in the s - vertex sequence of @xmath625 it follows from proposition [ prop : parityinv ] that if @xmath215 is direct then @xmath626 is even and part ( i ) of the conclusion follows . if @xmath215 is twisted then @xmath626 is odd , @xmath627 and ( ii ) follows .
having dealt in proposition [ prop : parity ] with the behaviour of parity under projection we next ask what can be inferred about s - cycles in @xmath628}$ ] from their projections . [ prop : stoichproj ] let @xmath215 be a cycle of @xmath628}$ ] .
\(i ) if @xmath215 is twisted then @xmath563 is an s - walk if and only if @xmath215 is an s - cycle .
\(ii ) if @xmath215 is direct , @xmath571 , and both @xmath514 and @xmath515 are s - walks , then @xmath215 is an s - cycle .
we may assume that @xmath215 is directed ; otherwise , if all its edges are undirected , we simply choose one of the two possible orientations . by definition
, @xmath215 is an s - cycle if @xmath629 since an s - to - r edge of @xmath628}$ ] projects to an s - to - r edge of @xmath186 with the same label , one has @xmath630 in case ( i ) , and @xmath631 in case ( ii ) .
similar equalities hold for r - to - s edges and the conclusion is now immediate from ( [ eq : sr = rs ] ) . in conjunction with lemma [ lem : swalks ] , proposition [ prop : stoichproj ] has an immediate notable consequence : [ cor : scycles[2 ] ] if @xmath186 is steady , then the same is true for @xmath628}.$ ] in particular , if @xmath186 is acyclic , then @xmath628}$ ] is steady .
[ rem : allscycles ] the condition that @xmath186 is steady is often fulfilled in applications to crns .
corollary [ cor : scycles[2 ] ] shows that when this is the case , verifying that @xmath628}$ ] is odd@xmath223 boils down to ruling out odd intersections of e - cycles .
liftings of dsr cycles as explained in remark [ rem : pi ] , the projection map @xmath495 is not injective ; in particular different cycles in @xmath362}_{a , b}$ ] may project to the same closed walk in @xmath169 .
if @xmath204 is a cycle of @xmath169 then any cycle @xmath215 of @xmath362}_{a , b}$ ] that projects to @xmath204 will be called a * lifting * of @xmath632 by way of definition [ def : projcycle ] and remark [ rem : extendpi ] , either @xmath215 is twisted or , if @xmath571 then @xmath633 or @xmath634 . in the first case @xmath215 will be called an * internal lifting * of @xmath204 and in the second , @xmath215 is an * external lifting * of @xmath632 the next proposition sheds light on this terminology .
[ prop : cyclelift ] let @xmath204 be a cycle in @xmath169 with vertex sequence @xmath635 .
\(i ) a cycle @xmath215 in @xmath362}_{a , b}$ ] is an external lifting of @xmath204 if and only if there exists @xmath636 such that @xmath637 \(ii ) if @xmath215 is an internal lifting of @xmath204 then the s - vertices of @xmath215 are a subset of @xmath638 \(i ) in one direction we can easily check by computing @xmath563 that a cycle of the form @xmath215 is an external lifting of @xmath204 . on the other hand , suppose an external lifting of @xmath204 gives rise to a cycle whose vertex sequence is not of this form , implying that @xmath215 contains three consecutive s - vertices @xmath639 with no common element ( namely @xmath640 are all distinct ) .
it follows from the proof of proposition [ prop : proj ] that each of @xmath514 and @xmath515 must include a vertex from @xmath641 and a vertex from @xmath642 which means that each of @xmath514 and @xmath515 contains at least two s - vertices and therefore at least two edges .
this contradicts the fact that one of them must be empty .
\(ii ) since @xmath215 is twisted , @xmath643 . if @xmath369 is an s - vertex of @xmath215 then , by construction , either @xmath507 is an s - vertex of @xmath514 and @xmath508 is an s - vertex of @xmath515 , or vice versa .
the conclusion follows .
[ rem : liftparity ] it is clear from proposition [ prop : parity ] that if a cycle @xmath215 in @xmath362}_{a , b}$ ] is an external lifting of a cycle @xmath204 in @xmath169 , then @xmath644 , while if @xmath215 is an internal lifting of @xmath204 , then @xmath645 . in other words
internal liftings change parity , while external liftings preserve parity of cycles in @xmath169 .
[ rem : shortcycles ] a cycle @xmath215 in @xmath362}_{a , b}$ ] of length @xmath646 must be of the form @xmath647 , with @xmath648 and @xmath649 .
thus , by observation @xmath563 is @xmath650 , namely a cycle of length @xmath646 in @xmath169 . as @xmath215 is clearly direct
, it has the same parity as @xmath563 .
thus by proposition [ prop : cyclelift ] cycles of length @xmath646 in @xmath362}_{a , b}$ ] arise precisely as external liftings of cycles of length @xmath646 in @xmath169 .
the same remark applies to cycles of length @xmath123 , although these can not occur in @xmath169 or @xmath362}_{a , b}$ ] if @xmath379 .
we may now argue that @xmath651}$ ] in this example is odd directly from @xmath470 .
consider a cycle @xmath437 of @xmath651},$ ] and let @xmath514 and @xmath515 denote the two walks in @xmath470 that form its projection .
since there are @xmath652 s - vertices in @xmath651},$ ] the length of @xmath437 can not exceed @xmath386 .
notice that @xmath470 has two cycles : an even hamiltonian cycle @xmath215 , and an odd 4-cycle @xmath137 . in view of remark [ rem : ww](2 ) , neither @xmath514 nor @xmath515 can traverse back and forward consecutively an unoriented edge of @xmath470 , and @xmath653 must therefore consist of copies of @xmath215 and @xmath137 .
in fact , there is at most one copy of each of @xmath215 and @xmath137 in this union : two copies of @xmath215 would imply the contradiction @xmath654 while two copies of @xmath137 would imply that @xmath437 contains both liftings of @xmath137 allowed by remark [ rem : shortcycles ] ( where @xmath655 is @xmath221 and @xmath646 respectively ) , so @xmath437 can not be a cycle itself .
now suppose that @xmath656 has the same edges as @xmath657 ( counted with multiplicity ) .
then the directed path @xmath658 is traversed twice , and therefore both directed paths @xmath659 and @xmath660 must exist in @xmath437 .
further , @xmath661 must arise as the projection of one of the directed paths @xmath662 or @xmath663 .
however , clearly @xmath664 and @xmath665 can not both exist in @xmath437 and similarly @xmath666 and @xmath667 can not both exist in @xmath437 , a contradiction .
we conclude that @xmath437 is either a lifting of @xmath215 or a lifting of @xmath137 .
since all liftings of e - cycle @xmath137 are internal and all liftings of o - cycle @xmath215 are external , it follows , by remark [ rem : liftparity ] , that @xmath651}$ ] is odd .
[ rem : multilift ] although this could not occur in the previous example , in general the same cycle in @xmath169 may lift to multiple cycles of different parities in @xmath362}_{a , b}$ ] , by lifting both externally and internally .
this is illustrated in the following example ( edge labels have been omitted ) : [ scale=1.2 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.3 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; at ( -1.9,0 ) @xmath668 ; ( 12 ) at ( -1,1 ) @xmath386 ; ( 23 ) at ( 1,1 ) @xmath388 ; ( 25 ) at ( -1,-1 ) @xmath594 ; ( 24 ) at ( 1,-1 ) @xmath593 ; ( [ 12 ] ) at ( 0,1 ) @xmath382 edge[pre,- ] ( 12 ) edge [ pre,- ] ( 23 ) ; ( [ 22 ] ) at ( 1,0 ) @xmath462 edge[pre,-,dashed ] ( 23 ) edge [ pre,- ] ( 24 ) ; ( [ 32 ] ) at ( 0,-1 ) @xmath669 edge[pre,-,dashed ] ( 24 ) edge [ pre,-,dashed ] ( 25 ) ; ( [ 42 ] ) at ( -1,0 ) @xmath473 edge[pre,- ] ( 25 ) edge [ pre,- ] ( 12 ) ; at ( -1.9,0 ) @xmath670 ; ( 12 ) at ( -1,1 ) @xmath389 ; ( 23 ) at ( 1,1 ) @xmath391 ; ( 25 ) at ( -1,-1 ) @xmath592 ; ( 24 ) at ( 1,-1 ) @xmath387 ; ( [ 12 ] ) at ( 0,1 ) @xmath384 edge[pre,-,dashed ] ( 12 ) edge [ pre,- ] ( 23 ) ; ( [ 22 ] ) at ( 1,0 ) @xmath423 edge[pre,- ] ( 23 ) edge [ pre,-,dashed ] ( 24 ) ; ( [ 32 ] ) at ( 0,-1 ) @xmath420 edge[pre,-,dashed ] ( 24 ) edge [ pre,- ] ( 25 ) ; ( [ 42 ] ) at ( -1,0 ) @xmath671 edge[pre,- ] ( 25 ) edge [ pre,-,dashed ] ( 12 ) ; at ( -1.7,0 ) @xmath672 ; ( 12 ) at ( -1,1 ) @xmath192 ; ( 23 ) at ( 1,1 ) @xmath460 ; ( 25 ) at ( -1,-1 ) @xmath589 ; ( 24 ) at ( 1,-1 ) @xmath190 ; ( [ 12 ] ) at ( 0,1 ) @xmath193 edge[pre,- ] ( 12 ) edge [ pre,- ] ( 23 ) ; ( [ 22 ] ) at ( 1,0 ) @xmath419 edge[pre,- ] ( 23 ) edge [ pre,- ] ( 24 ) ; ( [ 32 ] ) at ( 0,-1 ) @xmath191 edge[pre,- ] ( 24 ) edge [ pre,- ] ( 25 ) ; ( [ 42 ] ) at ( -1,0 ) @xmath466 edge[pre,- ] ( 25 ) edge [ pre,- ] ( 12 ) ; @xmath204 is an e - cycle , as is its external lifting @xmath673 ( note that it has four inversions ) . on the other hand ,
the internal lifting @xmath674 is an o - cycle with one inversion .
further applications and limitations of the dsr@xmath44}$ ] graph [ sec : final ] we present some further results which illustrate use of the theory developed so far .
in particular we show how it is possible in some cases , via the theory in section [ sec : theory ] , to make claims about @xmath284}$ ] by examination of the dsr graph @xmath186 and without explicit construction of @xmath362}_{a , b}$ ] . when the dsr graph has three s - vertices or three r - vertices [ sec3sp ] given @xmath675 and @xmath380 , then @xmath362}_{a , b}$ ] has a particularly nice structure , illustrated _ generically _ in the figure below : [ scale=.85 , place/.style = circle , draw = black!30,fill = black!0 , thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=<,shorten < = 2pt,>=stealth,semithick , post/.style = shorten > = 1pt,>=stealth,semithick ] ( 12 ) at ( -1.732,-1 ) 12 ; ( 23 ) at ( 1.732,-1 ) 23 ; ( 13 ) at ( 0 , 2 ) 13 ; ( [ 13 ] ) at ( 1.732,1 ) @xmath384 edge [ dashed , bend right = 20 , post ] ( 13 ) edge [ dashed , bend left = 20 , post ] ( 23 ) ; ( [ 12 ] ) at ( 0,- 2 ) @xmath383 edge [ dashed , bend left = 20 , post ] ( 12 ) edge [ bend right = 20 , post ] ( 23 ) ; ( [ 11 ] ) at ( -1.732,1 ) @xmath382 edge [ bend right = 20 , post ] ( 12 ) edge [ bend left = 20 , post ] ( 13 ) ; ( [ 23 ] ) at ( 1.5 * 1.732,1.5 * 1 ) @xmath420 edge [ dashed , bend right = 30 , post ] ( 13 ) edge [ dashed , bend left = 30 , post ] ( 23 ) ; ( [ 22 ] ) at ( 1.5 * 0,- 1.5 * 2 ) @xmath421 edge [ dashed , bend left = 30 , post ] ( 12 ) edge [ bend right = 30 , post ] ( 23 ) ; ( [ 21 ] ) at ( -1.5 * 1.732,1.5 * 1 ) @xmath422 edge [ bend right = 30 , post ] ( 12 ) edge [ bend left = 30 , post ] ( 13 ) ; at ( 1.8 * 1.732,1.8 * 1 ) @xmath676 ; at ( 1.8 * 0,- 1.8 * 1.95 ) @xmath677 ; at ( -1.8 * 1.732,1.8 * 1 ) @xmath676 ; ( [ 33 ] ) at ( 2.1 * 1.732,2.1 * 1 ) @xmath678 edge [ dashed , bend right = 40 , post ] ( 13 ) edge [ dashed , bend left = 40 , post ] ( 23 ) ; ( [ 32 ] ) at ( 2.1 * 0,- 2.1 * 2 ) @xmath679 edge [ dashed , bend left = 40 , post ] ( 12 ) edge [ bend right = 40 , post ] ( 23 ) ; ( [ 31 ] ) at ( -2.1 * 1.732,2.1 * 1 ) @xmath680 edge [ bend right = 40 , post ] ( 12 ) edge [ bend left = 40 , post ] ( 13 ) ; depending on the particular entries of @xmath77 and @xmath681 some edges may be absent , some edges may become directed , some edges may change sign , and some edges may get the label @xmath682 nonetheless , the generic structure of @xmath362}_{a , b}$ ] is always the same and in particular , the degree of any r - vertex is at most 2 .
this fact implies that no odd intersection of cycles is possible ( see remark [ remoddintersect ] ) and checking that @xmath362}_{a , b}$ ] is odd@xmath223 amounts to showing that all its e - cycles are s - cycles .
this has a pleasant translation into a property of the dsr graph , as follows : [ thm:3sp ] let @xmath683 and @xmath380 .
if @xmath169 satisfies ( i ) all e - cycles of length 4 are s - cycles ; and ( ii ) all o - cycles of length 6 are s - cycles , then @xmath362}_{a , b}$ ] is odd@xmath223 , @xmath284}$ ] is a @xmath89-matrix , and the nonreal spectrum of @xmath120 does not intersect @xmath83 .
if @xmath362}_{a , b}$ ] is odd@xmath223 , then the remaining conclusions follow from theorem [ mainp0thm ] and lemma [ j2p0 ] .
since r - vertices of @xmath362}_{a , b}$ ] have degree at most 2 , in order to show that @xmath362}_{a , b}$ ] is odd@xmath223 we need only show that all e - cycles of @xmath362}_{a , b}$ ] are s - cycles .
as explained in remark [ rem : shortcycles ] , all 4-cycles of @xmath362}_{a , b}$ ] are external liftings of 4-cycles of @xmath169 , whose parity ( and edge - labels ) they inherit .
it follows from ( i ) that all e - cycles of length 4 in @xmath362}_{a , b}$ ] are s - cycles . on the other hand ,
any 6-cycle @xmath215 of @xmath362}_{a , b}$ ] includes all s - vertices of @xmath628}$ ] , namely @xmath684 and @xmath387 , and therefore has exactly one inversion .
thus @xmath215 is twisted ( proposition [ prop : parityinv ] ) , @xmath563 is a closed walk , and moreover ( propositions [ prop : parity ] and [ prop : stoichproj ] ) @xmath215 is an e - cycle if and only if @xmath563 is odd , and @xmath215 is an s - cycle if and only if @xmath563 is an s - walk .
we may distinguish three possibilities for a 6-cycle @xmath215 , according to the three possible structures of its projection , illustrated below . here
vertices depicted as distinct are distinct , and the projected closed walks are obtained by following the arrows . as the hypothesis that @xmath380 means that no pair of vertices in @xmath169 are connected by more than one edge , two arrows between a pair of vertices mean that an undirected edge is traversed twice : [ scale=1.2 , place/.style = circle , draw = blue!50,fill = blue!5,thick , inner sep=0pt , minimum size=5.5 mm , transition/.style = rectangle , draw = black!50,fill = black!5,thick , inner sep=0pt , minimum size=5 mm , pre/.style=->,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; at ( -1.6,1.4 ) @xmath685 ; ( r1 ) [ transition , draw = black!30 , fill = black!0 ] at ( 30:1.2 ) @xmath686 ; ( r2 ) [ transition , draw = black!30 , fill = black!0 ] at ( -90:1.2)@xmath686 ; ( r3 ) [ transition , draw = black!30 , fill = black!0 ] at ( 150:1.2 ) @xmath686 ; ( 12 ) at ( -30:1.2 ) @xmath504 ; ( 23 ) at ( -150:1.2 ) @xmath504 ; ( 25 ) at ( 90:1.2 ) @xmath504 ; ( r1 ) to ( 12 ) ; ( 12 ) to(r2 ) ; ( r2 ) to ( 23 ) ; ( 23 ) to ( r3 ) ; ( r3 ) to ( 25 ) ; ( 25 ) to ( r1 ) ; at ( -1.6,1.4 ) @xmath687 ; at ( 0,0 ) @xmath686 ; ( r2 ) [ transition , draw = black!30 , fill = black!0 ] at ( -90:1.2 ) @xmath686 ; ( 12 ) at ( -30:1.2 ) @xmath504 ; ( 23 ) at ( -150:1.2 ) @xmath504 ; ( 25 ) at ( 90:1.2 ) @xmath504 ; ( 25 ) to ( 30:.25 ) ; ( 30:.25 ) to(12 ) ; ( 12 ) to ( r2 ) ; ( r2 ) to ( 23 ) ; ( 23 ) to ( 150:.25 ) ; ( 150:.25 ) to ( 25 ) ; at ( -1.6,1.4 ) @xmath688 ; at ( 0,0 ) @xmath686 ; ( 12 ) at ( -30:1.2 ) @xmath504 ; ( 23 ) at ( -150:1.2 ) @xmath504 ; ( 25 ) at ( 90:1.2 ) @xmath504 ; ( 25 ) to(30:.25 ) ; ( 30:.25 ) to(12 ) ; ( 12 ) to ( -90:.25 ) ; ( -90:.25 ) to ( 23 ) ; ( 23 ) to ( 150:.25 ) ; ( 150:.25 ) to ( 25 ) ; we now need to show that the hypotheses of the theorem imply that odd walks in @xmath169 of types ( a ) , ( b ) and ( c ) are closed s - walks .
type ( a ) is an acyclic walk and is structurally a closed s - walk ( see corollary [ cor : tree - s ] ) .
type ( b ) is composed of an undirected edge traversed twice and a 4-cycle .
if the walk is odd then the 4-cycle must be even and thus an s - cycle by hypothesis ( i ) ; it follows that the walk itself is a closed s - walk .
finally , type ( c ) is covered by hypothesis ( ii ) .
as we might expect , there is a dual result to theorem [ thm:3sp ] for matrices with three columns : [ thm:3reac ] let @xmath689 and @xmath380 .
if @xmath169 satisfies ( i ) all e - cycles of length 4 are s - cycles and ( ii ) all o - cycles of length 6 are s - cycles , then the nonreal spectrum of @xmath120 does not intersect @xmath83 .
observe that @xmath169 satisfies conditions ( i ) and ( ii ) on its cycles if and only if @xmath690 does . by theorem [ thm:3sp ] , the nonreal spectrum of @xmath450 does not intersect @xmath83 . since @xmath120 has the same nonzero spectrum as @xmath450 ( lemma [ spectranposelemma ] ) , the result follows .
the conclusions in example [ ex3row ] and proposition [ 3sp1 ] are an immediate consequence of theorem [ thm:3sp ] .
when the dsr graph is steady with a vertex set of low degree whereas the results of section [ sec3sp ] dealt with matrices restricted by size , we now consider matrices of arbitrary dimension but with restrictions on the number of nonzero entries in rows / columns .
the following generalisation of proposition [ prop : switchrs ] is immediate : [ thm : alls ] let @xmath162 be such that @xmath691 is steady .
* if @xmath77 has no more than two nonzero entries in each column then , for any @xmath161 , @xmath120 and @xmath284}$ ] are @xmath89-matrices and so @xmath120 is positive semistable . *
if @xmath77 has no more than two nonzero entries in each row then , for any @xmath161 , @xmath120 is positive semistable .
@xmath691 is steady by assumption and @xmath692}$ ] is steady by corollary [ cor : scycles[2 ] ] .
the proof now proceeds exactly as that of proposition [ prop : switchrs ] .
when the dsr graph is acyclic [ thm : tree ] let @xmath693 be such that @xmath694 is acyclic .
then , for each @xmath695 and @xmath696 , @xmath120 and @xmath284}$ ] are @xmath89-matrices and @xmath120 is positive semistable .
since @xmath694 is acyclic , it is odd , and so by theorem [ mainp0thm ] @xmath120 is a @xmath89-matrix for each @xmath697 . observe that ( i ) given any @xmath698 , since @xmath691 is acyclic , @xmath362}_{a , a^t } = g_{\overline { \mathbf{l}}^a , ( \overline { \mathbf{l}}^a)^t}$ ] is steady by corollary [ cor : scycles[2 ] ] and ( ii ) given any @xmath699 , @xmath700 and @xmath701 . by lemma [ lem : scyclepmat ] ( i ) and
( ii ) imply that @xmath284 } = \overline { \mathbf{l}}^a\underline { \mathbf{l}}^b$ ] is a @xmath89-matrix for each @xmath702 and the result extends by closure to @xmath697 . since for each @xmath697 both @xmath120 and @xmath284}$ ] are @xmath89-matrices , the conclusion that @xmath120 is positive semistable now follows from lemma [ lemposstable0 ] .
[ rem : lemcolun ] lemma [ lem : columnvec ] is clearly a special case of theorem [ thm : tree ] .
if @xmath215 is a column vector then @xmath703 is in fact a star .
[ rem : notoddstar ] if @xmath694 is acyclic , this does not imply that @xmath362}_{c , c^t}$ ] is odd@xmath223 . while most of the examples presented have involved situations where the dsr@xmath44}$ ] graph is odd or odd@xmath223 , theorem [ thm : tree ] illustrates that the dsr@xmath44}$ ] graph may provide useful information even in the case where it fails to be odd@xmath223 .
[ rem : notacyclic ] observe that theorem [ thm : tree ] does not imply that if @xmath186 is acyclic , then @xmath120 is positive semistable .
however , given @xmath704 , there exists @xmath705 with @xmath694 acyclic and such that @xmath695 , @xmath696 if and only if the underlying undirected graph of @xmath186 is acyclic ( i.e. , @xmath186 has no semicycles ) .
theorem [ thm : tree ] could thus be rephrased : `` if the underlying undirected graph of @xmath186 is acyclic , then @xmath120 and @xmath284}$ ] are @xmath89-matrices and @xmath120 is positive semistable . ''
there are a number of ways in which naive application of the theory described here can fail to provide information , even when a set of matrices structurally forbids the passage of a pair of nonreal eigenvalues through the imaginary axis .
we illustrate with a couple of examples . following up on remark [ rem : notoddstar ]
, the next example illustrates that even when @xmath362}_{a , b}$ ] is not odd@xmath223 and in fact @xmath284}$ ] is not necessarily a @xmath89-matrix , it may still be possible to infer the absence of hopf bifurcation from the dsr@xmath44}$ ] graph : [ exmatchings ] consider the matrices @xmath706 with @xmath707 .
@xmath362}_{a , b}$ ] is shown : [ scale=0.8 , place/.style = circle , draw = black!30,fill = black!0 , thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style=<,shorten < = 2pt,>=stealth,semithick , post/.style = shorten > = 1pt,>=stealth,semithick ] ( 12 ) at ( -1.732,-1 ) @xmath386 ; ( 23 ) at ( 1.732,-1 ) @xmath389 ; ( 13 ) at ( 0 , 2 ) @xmath387 ; ( [ 13 ] ) at ( 1.732,1 ) @xmath384 edge [ - > , dashed , bend right = 20 , post ] ( 13 ) edge [ < - , bend left = 20 , post ] ( 23 ) ; ( [ 12 ] ) at ( 0,- 2 ) @xmath383 edge [ - > , dashed , bend left = 20 , post ] ( 12 ) ; ( [ 11 ] ) at ( -1.732,1 ) @xmath382 edge [ < - , dashed , bend right = 20 , post ] ( 12 ) ; ( [ 23 ] ) at ( 1.5 * 1.732,1.5 * 1 ) @xmath420 edge [ < - , bend right = 30 , post ] ( 13 ) edge [ - > , dashed , bend left = 30 , post ] ( 23 ) ; ( [ 22 ] ) at ( 1.5 * 0,- 1.5 * 2 ) @xmath421 edge [ < - , bend left = 30 , post ] ( 12 ) edge [ < - , bend right = 30 , post ] ( 23 ) ; ( [ 21 ] ) at ( -1.5 * 1.732,1.5 * 1 ) @xmath422 edge [ - > , bend right = 30 , post ] ( 12 ) edge [ < - , bend left = 30 , post ] ( 13 ) ; ( [ 33 ] ) at ( 2 * 1.732,2 * 1 ) @xmath423 edge [ < - , dashed , bend right = 40 , post ] ( 13 ) ; ( [ 32 ] ) at ( 2 * 0,- 2 * 2 ) @xmath424 edge [ < - , dashed , bend left = 40 , post ] ( 12 ) edge [ bend right = 40 , post ] ( 23 ) ; ( [ 31 ] ) at ( -2 * 1.732,2 * 1 ) @xmath425 edge [ bend left = 40 , post ] ( 13 ) ; labels have been omitted , but it is easy to check that @xmath362}_{a , b}$ ] fails to be odd@xmath223 as a consequence of the cycle @xmath708 which is even but , involving @xmath185 labels , not an s - cycle .
indeed @xmath284}$ ] is not necessarily a @xmath89-matrix for all values of the entries .
however computation reveals that @xmath284}$ ] is nonsingular and applying lemma [ j2nonsingular ] , @xmath709 does not intersect the imaginary axis , forbidding hopf bifurcation .
in fact , more careful application of the theory developed in @xcite allows us to infer from @xmath362}_{a , b}$ ] , and without direct computation of @xmath710})$ ] , that @xmath710 } ) > 0 $ ] : the argument is only sketched here .
there are exactly two cycles in @xmath362}_{a , b}$ ] , an odd cycle @xmath711 and the even cycle @xmath215 .
however @xmath215 does not contribute any negative terms to the determinant : this follows from the fact that no undirected edges are incident into s - vertex @xmath386 and consequently there is no set of three r - vertices matched with the three s - vertices in both @xmath362}_{a , 0}$ ] and @xmath362}_{0 , b}$ ] in such a way that the union of these matchings includes @xmath215 .
see remark [ rem : p0proof ] ; see also the discussion on `` good '' cycles @xcite ( resp .
`` central '' cycles in @xcite ) and pfaffian orientations .
since @xmath215 is the only even cycle in @xmath362}_{a , b}$ ] and contributes nothing to the determinant , @xmath710 } ) \geq 0 $ ] .
that @xmath710 } ) \neq 0 $ ] follows from the existence of @xmath137 , which traverses each s - vertex and thus contributes a nonzero term to @xmath710})$ ] .
[ rem : adddiag1 ] remark [ rem : adddiag ] does not apply to example [ exmatchings ] , since @xmath284}$ ] has not been proved to be a @xmath89-matrix .
the next example demonstrates that @xmath284}$ ] may be a @xmath89-matrix for all @xmath161 although this does not follow from examination of @xmath362}_{a , a^t}$ ] using any of the theory presented so far .
it seems likely that this information can indeed be inferred from @xmath362}_{a , a^t}$ ] , but that new techniques will be needed .
[ ex4by5 ] consider the matrix @xmath712 symbolic computation confirms that @xmath284}$ ] is a @xmath89-matrix for arbitrary @xmath161 . since @xmath228 is odd , @xmath120 is also a @xmath89-matrix , and so , by lemma [ lemposstable0 ] , @xmath120 is in fact positive semistable .
however , neither @xmath362}_{a , a^t}$ ] nor @xmath362}_{a^t , a}$ ] is odd@xmath223 .
the picture on the left shows two e - cycles of @xmath362}_{a , a^t}$ ] with odd intersection . on the right @xmath362}_{\tilde a^t , \tilde a}$ ]
is depicted , where @xmath285 denotes the submatrix of @xmath77 obtained by removing its first and last rows .
the e - cycles @xmath713 and @xmath714 have odd intersection , and @xmath362}_{\tilde a^t , \tilde a}$ ] fails to be odd@xmath223 . by lemma [ lem : leafremove ] , @xmath362}_{a^t , a}$ ] is not odd@xmath223 either .
[ scale=2.1 , place/.style = circle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5.3 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , place1/.style = circle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5.5 mm , transition1/.style = rectangle , draw = black!0,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style = shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres/.style=->,shorten < = -1pt , shorten > = -1pt,>=stealth,semithick , pre1/.style=<-,shorten < = 1pt , shorten > = 1pt,>=stealth,semithick , pres1/.style=<-,shorten
< = -1pt , shorten > = 0pt,>=stealth,semithick ] ; ( 15 ) at ( 30:1 ) @xmath594 ; ( 14 ) at ( 90:1 ) @xmath388 ; ( 24 ) at ( 150:1 ) @xmath390 ; ( 34 ) at ( 210:1 ) @xmath391 ; ( 35 ) at ( 270:1 ) @xmath592 ; ( 45 ) at ( 330:1 ) @xmath593 ; ( 25 ) at ( -10:.4 ) @xmath578 ; ( [ 42 ] ) at ( 160:.4 ) @xmath472 edge[pre , bend right=7 , dashed , semithick ] ( 25 ) edge[pre , bend left=7,semithick ] ( 24 ) ; ( [ 15 ] ) at ( 0:1 ) @xmath715 edge[pre , bend right=10 , dashed , semithick ] ( 15 ) edge[pre , bend left=7,semithick ] ( 25 ) edge[pre , bend left=10,semithick ] ( 45 ) ; ( [ 41 ] ) at ( 60:1 ) @xmath473 edge[pre , bend right=10,semithick ] ( 14 ) edge[pre , bend left=10 , dashed , semithick ] ( 15 ) ; ( [ 14 ] ) at ( 120:1 ) @xmath385 edge[pre , bend right=10,semithick ] ( 24 ) edge[pre , bend left=10 , dashed , semithick ] ( 14 ) ; ( [ 24 ] ) at ( 180:1 ) @xmath462 edge[pre , bend right=10,semithick ] ( 34 ) edge[pre , bend left=10 , dashed , semithick ] ( 24 ) ; ( [ 43 ] ) at ( 240:1 ) @xmath671 edge[pre , bend right=10 , dashed , semithick ] ( 35 ) edge[pre , bend left=10,semithick ] ( 34 ) ; ( [ 35 ] ) at ( 300:1 ) @xmath669 edge[pre , bend right=10,semithick ] ( 45 ) edge[pre , bend left=10 , dashed , semithick ] ( 35 ) ; [ scale=1.3 , place/.style = circle , draw = black!30,fill = black!0 , thick , inner sep=0pt , minimum size=5.3 mm , place1/.style = circle , draw = white!5,fill = white!9,thick , inner sep=0pt , minimum size=5.3 mm , transition1/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , transition/.style = rectangle , draw = black!30,fill = black!0,thick , inner sep=0pt , minimum size=5 mm , pre/.style = shorten > = 2pt,>=stealth,semithick , dashed , post/.style = shorten > = 1pt,>=stealth,semithick ] ( 1,1.5 ) ( 3.3,1.5)(3.3,0 ) ; ( 1,1.5 ) ( -1.25,1.5)(-1.25,.86 ) ; ( 1,-1.5 ) ( 3.3,-1.5)(3.3,0 ) ; ( 1,-1.5 ) ( -1.25,-1.5)(-1.25,-.86 ) ; ( 24 ) at ( 0:1 ) @xmath390 ; ( 14 ) at ( 120:1 ) @xmath388 ; ( 34 ) at ( 240:1 ) @xmath391 ; ( 23 ) at ( 120:1 ) @xmath389 ; ( 12 ) at ( 240:1 ) @xmath386 ; ( 13 ) at ( 0:1 ) @xmath387 ; ( [ 14 ] ) at ( 60:1 ) @xmath385 edge [ post , bend right=15,semithick ] ( 14 ) edge [ pre , bend left=15,semithick ] ( 24 ) ; ( [ 34 ] ) at ( 180:1 ) @xmath471 edge [ post , bend left=15,semithick ] ( 14 ) edge [ post , bend right=15,semithick ] ( 34 ) ; ( [ 24 ] ) at ( 300:1 ) @xmath462 edge [ post , bend right=15,semithick ] ( 24 ) edge [ pre , bend left=15,semithick ] ( 34 ) ; ( [ 32 ] ) at ( 180:1 ) @xmath424 edge [ post , bend right=15,semithick ] ( 12 ) edge [ post ] ( 24 ) edge [ pre , bend left=15 ] ( 23 ) ; ( [ 21 ] ) at ( 300:1 ) @xmath422 edge [ pre , bend left=15 ] ( 12 ) edge [ post , bend right=15 ] ( 13 ) ; ( [ 13 ] ) at ( 60:1 ) @xmath384 edge [ post , bend left=15 ] ( 13 ) edge [ pre , bend right=15 ] ( 23 ) ; ( [ 31 ] ) at ( 1,1.5 ) @xmath425 ; ( [ 33 ] ) at ( 1,-1.5 ) @xmath423 ; notice that although @xmath362}_{a , a^t}$ ] is steady , lemma [ lem : scyclepmat ] can not be applied since @xmath362}_{b^t , b}$ ] is not in general steady
. an interesting feature of this example is that some products of the form @xmath716\underline{\mathbf{l}}^b[\beta|\alpha]$ ] in ( [ eqcb2 ] ) contain negative terms , but all such negative terms are cancelled by positive terms arising from other products .
the last two examples demonstrate that dsr@xmath44}$ ] graphs have more to offer than is explored in this paper .
in particular a more complete analysis of projection and lifting may provide new techniques to rule out hopf bifurcation in families of matrices .
we conclude by remarking that criteria such as `` @xmath284}$ ] is a @xmath89-matrix '' or even `` @xmath710 } ) > 0 $ ] '' which forbid the passage of a real eigenvalue of @xmath284}$ ] through zero are sufficient but not necessary to preclude hopf bifurcation .
indeed a real eigenvalue of @xmath284}$ ] may pass through zero as a consequence of migration of eigenvalues of @xmath120 on the real axis without any possibility of hopf bifurcation .
this limitation may be circumvented by resorting to conditions that distinguish imaginary pairs of eigenvalues of @xmath120 from pairs of real eigenvalues summing to zero , for example based on subresultant criteria @xcite .
natural extensions of the work here involve exploiting such criteria , and also examining other graphs related to @xmath120 whose cycle structure contains information about hopf bifurcation .
the authors acknowledge support from leverhulme grant f/07 058/bu `` structural conditions for oscillation in chemical reaction networks '' .
this work was completed during casian pantea s stay as a research associate in the department of electrical and electronic engineering , imperial college london .
g. craciun , c. pantea , and e. sontag .
graph - theoretic characterizations of multistability and monotonicity for biochemical reaction networks . in h.
koeppl , d. densmore , g. setti , and m. di bernardo , editors , _ design and analysis of biomolecular circuits _ , pages 239357 .
springer , 2011 .
m. mincheva and m. r. roussel .
graph - theoretic methods for the analysis of chemical and biochemical networks , i. multistability and oscillations in ordinary differential equation models . , 55:6186 , 2007 . | we describe combinatorial approaches to the question of whether families of real matrices admit pairs of nonreal eigenvalues passing through the imaginary axis .
when the matrices arise as jacobian matrices in the study of dynamical systems , these conditions provide necessary conditions for hopf bifurcations to occur in parameterised families of such systems .
the techniques depend on the spectral properties of additive compound matrices : in particular , we associate with a product of matrices a signed , labelled digraph termed a @xmath0}$ ] graph , which encodes information about the second additive compound of this product .
a condition on the cycle structure of this digraph is shown to rule out the possibility of nonreal eigenvalues with positive real part .
the techniques developed are applied to systems of interacting elements termed `` interaction networks '' , of which networks of chemical reactions are a special case .
hopf bifurcation ; compound matrices ; interaction networks + * msc .
* 15a18 ; 15a75 ; 05c90 ; 34c23 ; 37c27 |
in this paper we describe a new dynamical instability in superfluids .
this `` two - stream '' instability is analogous to the kelvin - helmholtz instability @xcite .
its key distinguishing feature is that the two fluids are interpenetrating rather than in contact across an interface as in the standard scenario .
the two - stream instability is well known in plasma physics [ where it is sometimes referred to as the `` farley - buneman '' instability @xcite ] , and it has also been discussed in various astrophysical contexts like merging galaxies @xcite and pulsar magnetopheres @xcite , but as far as we are aware it has not been previously considered for superfluids . in fact , the `` standard '' kelvin - helmholtz instability was only recently discussed in the context of superfluids @xcite . the similarity of the equations used in plasma physics [ a nice pedagogical description of the plasma two - stream instability can be found in @xcite ] to the ones for two - fluid superfluid models inspired us to ask whether an analogous instability could be relevant for superfluids .
that this ought to be the case seemed inevitable . to prove the veracity of this expectation , we have adapted the arguments from the plasma problem to the superfluid case , and discuss various aspects of the two - stream instability in this paper . of particular interest to us
is the possibility that the two - stream instability may operate in rotating superfluid neutron stars .
mature neutron stars are expected to be sufficiently cold ( eg . below @xmath0 k ) that their interiors contain several superfluid / superconducting components .
such loosely coupled components are usually invoked to explain the enigmatic pulsar glitches , sudden spin - up events where the observed spin rate jumps by as much as one part in @xmath1 @xcite .
theoretical models for glitches @xcite have been discussed ever since the first vela pulsar glitch was observed in 1969 @xcite , but these event are still not well understood .
after three decades of theoretical effort it is generally accepted that the glitches arise because a superfluid component can rotate at a rate different from that of the bulk of the star , and that a transfer of angular momentum from the superfluid to the crust of the star could lead to the observed phenomenon .
the relaxation following the glitch is well explained in terms of vortex creep [ see for example @xcite ] , but the mechanism that triggers the glitch event remains elusive . in this context , it seems plausible that the superfluid two - stream instability may turn out to be relevant .
we take as our starting point the two - fluid equations that have been used to discuss the dynamics of superfluid neutron stars @xcite .
hence , we consider superfluid neutrons ( index @xmath2 ) coexisting with a conglomerate of charged components ( index @xmath3 ) .
the constituents of the latter ( mainly ions and electrons in the neutron star crust and protons and electrons in the core ) are expected to be coupled by viscosity and the magnetic field on a very short timescale .
hence , we assume that these charged components will move together and that it is appropriate to treat them as a single fluid .
the corresponding equations are @xcite @xmath4 where @xmath5 represent the respective number densities and @xmath6 are the two velocities . here , and in the following , we use the constituent index @xmath7 which can be either @xmath2 or @xmath3 .
the respective mass densities are obviously given by and we further introduce the relative velocity @xmath8 between the two fluids as @xmath9 the first law of thermodynamics is defined by the differential of the energy density or `` equation of state '' , , namely @xmath10 which defines the chemical potentials @xmath11 and the `` entrainment '' function @xmath12 as the thermodynamic conjugates to the densities and the relative velocity . with these definitions
we can write the two euler - type equations : @xmath13 where we have introduced the dimensionless entrainment parameters @xmath14 the equation for the gravitational potential @xmath15 is @xmath16 where @xmath17 .
when @xmath18 these equations make manifest the so - called entrainment effect .
the entrainment arises because the bare neutrons ( or protons ) are `` dressed '' by a polarization cloud of nucleons comprised of both neutrons and protons .
since both types of nucleon contribute to the cloud the momentum of the neutrons , say , is modified so that it is a linear combination of the neutron and proton particle number density currents ( the same is true for the proton momentum ) .
this means that when one of the fluids begins to flow it will , through entrainment , induce a momentum in the other .
because of entrainment a portion of the protons ( and electrons ) will be pulled along with the superfluid neutrons that surround the vortices by means of which the superfluid mimics bulk rotation .
this motion leads to magnetic fields being attached to the vortices , and dissipative scattering of electrons off of these magnetic fields .
this `` mutual friction '' is expected to provide one of the main dissipative mechanisms in superfluid neutron star cores @xcite . in order to establish the existence of the superfluid two - stream instability we consider the following model problem : let the unperturbed configuration be such that the `` protons '' remain at rest , while the neutrons flow with a constant velocity @xmath19 . for simplicity
, we neglect the coupling through entrainment , i.e. we take @xmath20 , and we also neglect perturbations in the gravitational potential . under these assumptions ,
the two fluids are only coupled `` chemically '' through the equation of state .
writing the two velocities as @xmath21\hat{x}$ ] and @xmath22 where @xmath23 and @xmath24 are taken to be suitably small , we get the perturbation equations @xmath25 and @xmath26 next , we assume harmonic dependence on both @xmath27 and @xmath28 , i.e. we use the fourier decomposition @xmath29 $ ] etcetera .
this leads to the four equations @xmath30 we thus have four equations relating the six unknown variables @xmath31 , @xmath32 and @xmath33 . to close the system we need to provide an equation of state . given an energy functional @xmath34 we have @xmath35 and similarly @xmath36 finally , we define the two sound speeds by , cf .
@xcite , @xmath37 and introduce the `` coupling parameter '' @xmath38 which also has the dimension of a velocity squared .
for later convenience we have given the relation to the coefficients of the `` structure matrix '' @xmath39 used by @xcite .
with these definitions we get @xmath40 and we can rewrite our set of equations as @xmath41 \ , \\
n_\p \delta v_\p & = & { \omega \over k } \delta n_\p = { k \over \omega } \left [ { n_\p \over n_\n } { \cal c } \delta n_\n + c_{\p}^2 \delta n_\p\right ] \ . \end{aligned}\ ] ] reshuffling we get @xmath42 \delta n_\n & = & { \cal c } \delta n_\p \ , \\ \left [ \left ( { \omega \over k}\right)^2 - c_\p^2 \right]\delta n_\p & = & { n_\p \over n_\n } { \cal c } \delta n_\n \ , \end{aligned}\ ] ] and a dispersion relation @xmath43\left [ \left ( { \omega \over k}\right)^2 - c_\p^2 \right ] = { n_\p \over n_\n } { \cal c}^2 \ .\ ] ] introducing the `` pattern speed '' ( the phase velocity ) @xmath44 we have @xmath45 \left [ \pat^2 - c_\p^2 \right ] = { n_\p \over n_\n } { \cal c}^2 \ . \label{disp0}\ ] ] not surprisingly , this local dispersion relation is qualitatively similar to the one for the plasma problem @xcite
. we will now use it to investigate under what circumstances we can have complex roots , i.e. a dynamical instability .
first of all , it is easy to see that ( [ disp0 ] ) leads to the simple roots @xmath46 in the uncoupled case , when @xmath47 .
this establishes the interpretation of @xmath48 as the sound speeds . to investigate the coupled case
, we introduce new variables @xmath49 and @xmath50 .
then we get @xmath51[x^2-b^2 ] = 1 \label{fxy}\ ] ] where we have defined @xmath52 the onset of dynamical instability typically corresponds to the merger of two real - frequency modes
. if this is the case , a marginally stable configuration will be such that ( [ fxy ] ) has a double root .
this happens when an inflexion point of @xmath53 coincides with @xmath54 .
this is a useful criterion for searching for the marginally stable modes of our system . as a `` proof of principle '' we consider the particular case of @xmath55 and @xmath56 ( we will motivate this particular choice in section iie ) .
the real and imaginary parts of the mode - frequencies for these parameter values are shown as functions of @xmath57 in figure [ modes ] .
we have complex roots ( an instability ) in the range @xmath58 .
the corresponding mode frequencies lie in the range @xmath59 .
the fastest growing instability occurs for @xmath60 for which we find that @xmath61 .
in other words , in this particular case we encounter the two - stream instability once the rate of the background flow is increased beyond @xmath62 the corresponding frequency is given by @xmath63 from this we see that the instability is present well before the neutron flow becomes `` supersonic '' . this is crucial since one would expect the superfluidity to be destroyed for supersonic flows . ) for model parameters @xmath55 and @xmath56 . for these parameters
the quartic dispersion relation has four real roots for both @xmath64 and @xmath65 , while it has two real roots and a complex conjugate pair for @xmath57 in the range @xmath66 . in this range ,
the two - stream instability is operating .
the grey area corresponds to @xmath67 which is contained in the range of the instability criteria discussed in section iic .
, height=226 ] we have thus established that the two - stream instability may , in principle , operate in superfluids .
our example indicates the existence of a lower limit of the background flow for the instability .
this turns out to be a generic feature .
in contrast , the plasma two - stream instability can operate at arbitrarily slow flows .
an ideal plasma is unstable to sufficiently long wavelengths for any given @xmath19 . in reality , however , one must also account for dissipative mechanisms . in the case of real plasmas
one finds that the so - called landau damping stabilizes the longest wavelengths @xcite .
thus the two - stream instability sets in below a critical wavelength in more realistic plasma models , and there is typically ( just like in the present case ) a range of flows for which the instability is present .
we will discuss the effects of dissipation on the superfluid two - stream instability briefly in section iv .
it is useful to consider whether we can derive a necessary condition for the two - stream instability.to approach this problem in full generality would likely be quite complicated , but we can make good progress for the simple one - dimensional toy problem discussed above .
we begin by multiplying the euler equation ( [ eul_n ] ) for the neutrons by the complex conjugate @xmath68 .
this leads to ( after also using the continuity equations to replace the perturbed number densities ) @xmath69 similarly , we obtain from the second euler equation ( [ eul_p ] ) @xmath70 next we combine these two equations to get @xmath71 where we have introduced the pattern speed @xmath72 . from this expression
we see that the imaginary part of the left - hand side must vanish , so we should have i m @xmath73 .
allowing the pattern speed to be complex , i.e. using @xmath74 we find that the following condition must be satisfied @xmath75 | \delta v_\n|^2 + \left [ 2 - { v_0 \over \sigma_r } \left ( 1 + { c_\p^2 \over |\pat|^2 } \right)\right ] mode , @xmath76 , the frequency clearly must be such that the factors multiplying the absolute values of the two velocities have different signs .
let us first consider the case when the factor multiplying @xmath77 is negative .
then we find that an instability is only possible if @xmath78 , and the following condition is satisfied : @xmath79 this shows that we must have @xmath80 which constrains the permissible frequencies to the range @xmath81 .
thus we see that the flow must be subsonic , i.e. @xmath82 . in the case when the factor multiplying @xmath83 is negative we can only have an instability if @xmath84 .
we also require @xmath85 or @xmath86 , the condition that must be satisfied is ( [ cond1 ] ) .
it is useful to notice two things about this criterion .
first of all , any unstable mode for which @xmath87 must lie in the range @xmath88 . secondly , when @xmath89 the permissible range will be well approximated by @xmath90 , cf .
figure [ crit ] . as is clear from figure [ modes ] the unstable modes satisfy this last , and most severe , criterion . ) which is relevant for the example considered in section iib .
this example is constructed by introducing @xmath91 , and then showing the four curves : @xmath92 , @xmath93 , @xmath94 and @xmath95 .
here we have taken @xmath96 and @xmath97 . criterion ( [ cond1 ] )
is satisfied when @xmath98 and @xmath99 ( in the grey area ) .
the corresponding range is well approximated by @xmath100 .
, height=226 ] it is worth noting that the instability can be discussed in terms of a simple energy argument [ see @xcite and @xcite for similar discussions in other contexts ] .
after averaging over several wavelengths , the kinetic energy of the protons is @xmath101 meanwhile we get for the neutrons @xmath102 which leads to @xmath103\ ] ] from which we see that the energy in the perturbed flow is smaller than the energy in the unperturbed case , which means that we can associate the wave with a `` negative energy '' , when @xmath104 a wave that satisfies @xmath105 moves forwards with respect to the protons but backwards according to an observer riding along with the unperturbed neutron flow . as we have seen above , the unstable modes in our problem satisfy this criterion and hence it is easy to explain the physical conditions required for the two - stream instability to be present .
having established that the two - stream instability may be present in superfluids , we want to assess to what extent one should expect this mechanism to play a role for astrophysical neutron stars . to do this
we will consider two particular equations of state .
the results we obtain illustrate different facets of what we expect to be a rich problem .
we begin by making contact with our recent analysis of rotating superfluid models @xcite as well as the study of oscillating non - rotating stars by @xcite .
from the definitions above we have @xmath106 we combine these results with the explicit structure matrix given in eq .
( 144 ) of @xcite , which is based on a simple `` analytic '' equation of state .
this leads to @xmath107 where @xmath108 is the proton fraction , and @xmath109 is defined by @xmath110 as discussed by @xcite , @xmath109 is related to the `` symmetry energy '' of the equation of state , cf .
@xcite . ) . for slower flows ,
these critical curves approach the absolute instability regions .
the region where the two - stream instability may operate in physical flows therefore lies between each dashed curve and the nearest grey area . for comparison
we also indicate the curve in the @xmath111 plane traced out by the pal equation of state ( discussed in section iie ) as the density is varied from that near the crust / core interface ( @xmath112 ) to five times that of nuclear saturation @xmath113.,height=226 ] the instability regions for this model equation of state are illustrated in figure [ domain ] .
a key feature of this figure is the presence of regions of `` absolute instability '' .
this happens when @xmath114 .
then there exist unstable solutions already for vanishing background flow , @xmath64 . that this is the case is easy to see .
consider @xmath115 in the limit @xmath64 . in the limit we can solve directly for @xmath116 : @xmath117
from which it is easy to see that one of the roots for @xmath116 will be negative if @xmath114 .
hence , one of the roots to the quartic ( [ fxy ] ) will be purely imaginary .
the physics of this instability is quite different from the two - stream instability that is the main focus of this paper .
yet it is an interesting phenomenon . from the above relations we find that @xmath118 corresponds to @xmath119 in the discussion by @xcite it was assumed that `` reasonable '' equations of state ought to satisfy this condition .
we expected this to be the case since the structure matrix would not be invertible if its determinant were to vanish at some point .
we now see that this constraint has a strong physical motivation : the condition is violated when @xmath114 , i.e. when we have an absolute instability .
the regions where this instability is active are indicated by the grey areas in figure [ domain ] . in order to strengthen the argument that the two - stream instability may operate in astrophysical neutron stars ,
we have considered a `` realistic '' equation of state due to prakash , ainsworth and lattimer ( pal ) ( 1988 ) .
the advantage of this model is that it is relatively simple . in particular
, it leads to analytical expressions for the various quantities needed in our analysis .
the energy density of the baryons for the pal equation of state can be written @xmath120 \ , \ ] ] where @xmath121 corresponds to the energy per nucleon , @xmath122 corresponds to the `` symmetry energy '' ( and is closely related to @xmath109 in the `` analytic '' equation of state discussed above ) , and @xmath123 with @xmath124 the nuclear saturation density .
@xmath121 takes the following form : @xmath125 \ , \ ] ] with @xmath126 mev , @xmath127 mev , @xmath128 mev , @xmath129 , @xmath130 mev , @xmath131 mev , @xmath132 , and @xmath133 .
the symmetry term is @xmath134 + s_0 f(u ) \ , \ ] ] with @xmath135 mev and @xmath136 mev . here
@xmath137 is a function satisfying @xmath138 which is supposed to simulate the behaviour of the potentials used in theoretical calculations . in our study
we have only considered @xmath139 , which is one of four possibilities discussed by @xcite .
we further need to account for the energy contribution of the electrons , which is important since the electrons are highly relativistic inside neutron stars .
hence , they can obtain high ( local ) fermi energies which may be comparable with the proton ( local ) fermi energies . considering only the electrons ,
the leptonic contribution to the energy density is given by ( in units where the speed of light is unity ) @xcite @xmath140 where @xmath141 is the electron mass ( in terms of the baryon mass @xmath142 ) , @xmath143 is the electron compton wavelength , and @xmath144^{1/2 } \left[1 + 2 x^2\right ] - { \rm ln}\left[x + \left(1 + x^2 \right)^{1/2}\right]\right ) \ , \ ] ] @xmath145 ^ 3 \right)^{1/3 } n_\p^{1/3 } \ .\ ] ] in doing this calculation we have assumed local charge neutrality , i.e. @xmath146 . the above energy term is added linearly in the equation of state .
having obtained an expression for the total energy as a function of the density , we can derive explicit expressions for all quantities needed to discuss the two - stream instability .
first we need to determine the proton fraction @xmath147 .
we do this by assuming that the star is in chemical equilibrium , i.e. @xmath148 solving ( [ chemeq ] ) for @xmath147 provides us with the proton fraction as a function of @xmath149 . given this , and the relevant partial derivatives of @xmath150 we can readily evaluate the symmetry energy and well as the sound speeds @xmath151 , @xmath152 and the chemical coupling parameter @xmath153 . with this data we can determine the two parameters @xmath154 and
@xmath155 which are needed if we want to solve the local dispersion relation ( [ fxy ] ) .
the results we obtain for the proton fraction and the symmetry energy are indicated in figure [ domain ] .
we consider the range from @xmath112 , presumed to correspond to the core - crust boundary , to @xmath156 which represents the deep core of a realistic neutron star .
the corresponding results for the two - stream instability are shown in figure [ local ] . from this figure
we can see that the two - stream instability may operate ( albeit at comparatively large relative flows ) in the region immediately below the crust . finally , we find that the conditions at the core - crust interface are such that @xmath55 and @xmath56 .
these are the values we chose for the example in section iib and hence the results shown in figure [ modes ] correspond to a physically realistic model . .
we indicate the location of the core - crust boundary ( @xmath157 ) by a vertical dashed line .
our model is only relevant for the core fluid , i.e. to the right of the vertical line .
finally , the horizontal dashed line indicates when the relative flow is equal to the ( neutron ) sound speed .
we expect the superfluid degeneracy to be broken beyond this level , so an instability located above this line is unlikely to have physical relevance .
the results indicate that there may be a region of instability immediately below the crust .
right panel : the corresponding oscillation frequencies . particularly notable is the point near @xmath158 where the two critical curves cross . at this point
the symmetry energy @xmath109 changes sign , cf .
[ domain ] , and there exist a particular density such that the two fluids are uncoupled , cf .
( [ adef ] ) .
, height=226 ]
the fact that a superfluid neutron star may , in principle , exhibit the two - stream instability does not necessarily prove that this mechanism will be astrophysically relevant .
yet , it is an intriguing possibility given that the mechanism underlying the enigmatic glitches observed in dozens of pulsars remains poorly understood .
one plausible astrophysical role for the superfluid two - stream instability would be in this context : perhaps this instability serves as trigger mechanism for large pulsar glitches ? our aim in this section is to construct a toy problem that allows us to investigate a possible connection between the two - stream instability and pulsar glitches .
a suitably simple problem corresponds to two fluids , allowed to rotate at different rates , confined within an infinitesimally thin spherical shell . by assuming that the shell is infinitesimal we ignore radial motion ,
i.e. we restrict the permissible perturbations of this system is such a way that the perturbed velocities must take the form @xmath159 where @xmath160 are the spherical harmonics and @xmath161 is the radius @xmath161 of the shell .
this means that the system permits only toroidal mode - solutions .
in other words , all oscillation modes of this shell model are closely related to the inertial r - modes of rotating single fluid objects @xcite .
the perturbation equations for the configuration we consider have been derived in a different context and the complete calculation will be presented elsewhere @xcite .
our primary interest here concerns whether the modes of this system may suffer the two - stream instability .
the presence of the instability in this toy problem would be a strong indication that it will also be relevant when the shells are `` thick '' and radial motion is possible .
that is , when we consider a rotating star that contains a partially decoupled superfluid either in the inner crust or the core .
as discussed in section iic , the two - stream instability can be understood in terms of negative energy waves . in the current problem ,
the criterion for waves to carry negative energy according to one fluid but positive energy according to the other fluid is that the mode pattern speed [ we are assuming a decomposition @xmath162 @xmath163 lies between the two ( uniform ) rotation rates . in other words , a necessary condition for instability is @xmath164 where we have assumed that the superfluid neutrons lag behind the charged component , as is expected in a spinning down pulsar . after a somewhat laborious calculation , see @xcite for details , one finds that the dispersion relation for the toroidal two - fluid modes of the shell problem is @xmath165 + 2 m \omega_\n + m \eps_\n \left[2 - l \left(l + 1\right)\right ] \left[\omega_\p - \omega_\n\right]\right\}\cr & & \times \left\ { - l \left(l + 1\right ) ( 1 - \eps_\p ) \left[\omega + m \omega_\p\right ] + 2 m \omega_\p + m \eps_\p \left[2 - l \left(l + 1\right)\right]\left[\omega_\n - \omega_\p\right]\right\ } \cr & & - \left\ { l \left(l + 1\right)\right)^2 \eps_\n \eps_\p \left(\omega + m \omega_\n\right )
\left(\omega + m \omega_\p\right\ } = 0 \ .\end{aligned}\ ] ] this equation should be valid under the conditions in the outer core of a mature neutron star , where superfluid neutrons are permeated by superconducting protons .
we rewrite this dispersion relation in terms of the entrainment parameter used by @xcite , namely @xmath166 the frequency as measured with respect to the rotation of the protons , @xmath167 and a dimensionless measure of the relative rotation , @xmath168 with these definitions we get @xmath169(\kappa+\y ) -2(1-x_\p)\y + x_\p \eps(l-1)(l+2 ) ( 1-\y ) \right\ } \nonumber\\ & & \times \left\ { l(l+1 ) [ 1-\eps](\kappa+1 ) -2 - \eps(l-1)(l+2 ) ( 1-\y ) \right\ } \nonumber \\ & & - [ l(l+1)]^2 x_\p \eps^2 ( \kappa+\y)(\kappa+1 ) = 0 \ .
\label{disper}\end{aligned}\ ] ] in terms of these new variables a mode would satisfy the necessary instability criterion ( [ nec_crit ] ) if @xmath170 is such that @xmath171 as we will now establish , there exist modes that satisfy this criterion for reasonable parameter values .
let us first consider the case of quadrupole oscillations , i.e. take @xmath172 .
typical results for this case are shown in figures [ shell ] and [ rotfreq ] .
the first figure illustrates the regions of the @xmath173 parameter space for which an instability is present in the case when ( i ) the neutrons rotate at a rate that is 90% faster than that of the charged component , and ( ii ) the neutrons lag behind by the same fraction .
the second figure shows the mode - frequencies corresponding to the second case [ simply obtained by solving the quadratic ( [ disper ] ) ] for the particular value @xmath174 .
this figure shows the presence of unstable modes within the range of values for @xmath175 that we take to be physically realistic @xcite : @xmath176 . from this figure
we immediately deduce two things .
first , we see that the unstable modes indeed satisfy the criterion ( [ yrange ] ) .
secondly , the unstable modes may have imaginary parts as large as i m @xmath177 .
it is useful to ask what this implies for the growth time of the instability .
since our results only depend on the azimuthal index @xmath178 through the scaling @xmath179 we see that the fastest growth time corresponds to the @xmath180 modes .
the e - folding time for the instability follows from @xmath181 where @xmath182 represents the observed rotation period of the pulsar ( presumably corresponding to the rotation of the charged component , i.e. @xmath183 ) . and relative rotation such that the neutrons rotate 90% slower ( left frame ) or faster ( right frame ) than the protons .
the two - stream instability operates in regions 1 - 2 .
the grey box corresponds to the `` physically reasonable '' part of parameter space ( for the core of a neutron star).,height=226 ] , @xmath184 and @xmath174 .
the corresponding instability region can be deduced from the right panel of fig .
[ shell ] .
the grey area in the left frame indicates the region where an instability is permissible according to ( [ yrange ] ) .
, height=226 ] the above example shows that the two - stream instability does indeed operate in this shell problem .
in fact , the analysis goes beyond the local analysis of the plane parallel problem in section iib since we have now solved for the actual unstable modes ( satisfying the relevant boundary conditions ) .
we see that , as expected for a dynamical instability , the growth time of an unstable mode may be very short . however , the relative rotation rates required to make the quadrupole modes unstable in the range @xmath185 and @xmath186 are likely far too large to be physically attainable . in this sense the results shown in figs .
[ shell][rotfreq ] are , despite being instructive , somewhat extreme .
a quantity of key importance for this discussion is the rotational lag between the two components . in order to be able to argue that the two - stream instability is relevant for pulsar glitches we need to consider lags that may actually occur in astrophysical neutron stars . to estimate the size of the rotational lag required to `` explain '' the observed glitches we assume that a glitch corresponds to a transfer of angular momentum from a partially decoupled superfluid component ( index @xmath2 ) to the bulk of the star ( index @xmath3 ) .
then we have @xmath187 where @xmath188 are the two moments of inertia .
now assume that the decoupled component corresponds 1% of the total moment of inertia , eg .
the superfluid neutrons in the inner crust or a corresponding amount of fluid in the core .
this would mean that @xmath189 , and we have @xmath190 combine this with the observations of large vela glitches to get @xmath191 in other words , we must have @xmath192 if we assume that the glitch brings the two fluids back into co - rotation , then we have @xmath193 and we see that the two rotation rates will maximally differ by one part in @xmath194 or so .
rotational lags of this order of magnitude have often been discussed in the context of glitches . even though the key quantity in models invoking catastrophic vortex unpinning in the inner crust the pinning strength
is very uncertain , and there have been suggestions that the pinning force is too weak to allow a build up of the required rotational lag @xcite , typical values considered are consistent with our rough estimate .
in addition , frictional heating due to a difference in the rotation rates of the bulk of a neutron star and a superfluid component has been discussed as a possible explanation for the fact that old isolated pulsars seem to be somewhat hotter than expected from standard cooling models @xcite .
@xcite argue that a lag of @xmath195 could explain the observational data . finally , the presence of rotational lags of the proposed magnitude is supported by a statistical analysis of 48 glitches in 18 pulsars @xcite .
this study suggests that the critical rotational lag at which a glitch occurs is @xmath196 in order to make our shell model problem more realistic we consider the case when the superfluid neutrons lag behind the superconducting protons , and take @xmath197 , or @xmath198 , as a representative value . with this rotational lag ,
we see from ( [ yrange ] ) that the unstable modes must be such that @xmath199 .
a series of results for this choice of parameters are shown in figures [ rotdomain]-[imag25 ] .
figure [ rotdomain ] illustrates the fact that , if we decrease the rotational lag then the two - stream instability will not be active ( in the interesting region of parameter space ) for low values of @xmath200 . for a smaller rotational
lag the instability acts on a shorter length scale .
for @xmath198 we find that we must have @xmath201 in order for there to be a region of instability in the part of the @xmath173 plane shown in figure [ rotdomain ] .
this means that the instability only operates on length scales shorter than @xmath202 m ( if we take the shell radius to be @xmath203 km ) . or @xmath198 .
we show results for four different values of @xmath200 .
there are no unstable modes in this part of parameter space unless @xmath201 .
recall that @xmath204 and @xmath176 would be reasonable parameter values for a neutron star core .
the grey diagonal line represents the singularity discussed in section iiid and the dashed curves indicate the instability regions in the large @xmath200 limit.,height=302 ] figure [ freq25 ] shows the real part of the mode frequencies for @xmath205 and various values of @xmath147 . from this figure
we can see that the instability always occur in the region suggested by the instability condition ( [ yrange ] ) , i.e. two real frequency modes never merge to give rise to a complex conjugate pair of solutions outside the grey areas indicated in the various panels of figure [ freq25 ] .
the imaginary parts for @xmath205 and the various values of @xmath147 considered in figure [ freq25 ] are shown in figure [ imag25 ] . from this figure
we see that the imaginary part of @xmath170 typically reaches values of order @xmath206 .
in fact , by comparing similar results for different values of @xmath200 we have found that the largest attainable imaginary part of @xmath170 varies by less than one order of magnitude as @xmath200 increases from 100 to 1000 .
thus we estimate that the typical instability growth time for @xmath207 will be @xmath208 for a star rotating at the rate of the vela pulsar , @xmath209 ms , we would have @xmath210 s for @xmath211 .
interestingly , this predicted growth time is significantly shorter than the resolved rise time of a large vela glitch @xmath212 s @xcite . ) as function of the entrainment parameter @xmath175 for @xmath205 , @xmath207 and various values of @xmath147 .
the range in which an instability is permissible [ according to ( [ yrange ] ) ] is indicated by the grey areas.,height=302 ] ( determining the growth rate ) of the unstable modes for @xmath207 and @xmath205 is shown as a function of @xmath175 for several different values of @xmath147 ( as indicated in the figure ) .
essentially , these results represent horizontal cuts through the frames in figure [ rotdomain ] .
, height=226 ] the results obtained above indicate that the two - stream instability is likely to act on modes with rather short wavelengths .
given this it makes sense to consider the large @xmath200 limit in more detail . keeping only the leading order term ( proportional to @xmath213 ) in ( [ disper ] ) we have the dispersion relation @xmath214\kappa \nonumber \\ & + & ( 1 - 2\eps - x_\p)\y + \eps x_\p + \eps(1-x_\p)\y^2 - 2x_\p\eps^2(1-\y)^2 = 0 \ . \end{aligned}\ ] ] we can readily write down the solutions to this quadratic : @xmath215^{1/2 } \right\ } \ .
\label{kappas}\end{aligned}\ ] ] it turns out that we can learn a lot about the problem from this expression .
the most obvious feature is the fact that @xmath216 have a singularity when @xmath217 .
it is straightforward to show that one of the roots will become infinite at this point , while the second root becomes : @xmath218 \ .\ ] ] for the examples shown in figure [ rotdomain ] this special case corresponds to the intersection between the instability domain and the diagonal line @xmath217 , as indicated in figure [ rotdomain ] .
it is also straightforward to deduce the regions of instability from the sign of the argument of the square - root .
we see that we will have an unstable mode when @xmath219 or @xmath220 these regions are also indicated in figure [ rotdomain ] . out of these two possible instability regions ,
the first is most likely to be relevant for neutron stars since it allows the instability to be present already for small proton fractions .
finally , we can use ( [ kappas ] ) to show the existence of an extremum at @xmath221 the corresponding imaginary part of @xmath170 would represent the fastest possible growth time for an unstable mode located in region 1 of figure [ rotdomain ] .
this leads to the estimate @xmath222 or an estimate of the shortest growth time : @xmath223 which agrees well with the result we previously obtained for @xmath211 .
hence , this simple formula can be used to estimate the fastest growth rate of the two - stream instability in our shell model for different parameter values .
in this paper we have introduced the superfluid two - stream instability : a dynamical instability analogous to that known to operate in plasmas @xcite , which sets in once the relative flow between the two components of the system reaches a critical level .
we have studied this instability for two different model problems .
first we analysed a local dispersion relation derived for the case of a background such that one fluid was at rest while the other had a constant flow rate .
this provided a proof of principle of the existence of the two - stream instability for superfluids .
our analysis was based on the two - fluid equations that have been used to model the dynamics of the outer core of a neutron star , where superfluid neutrons are expected to coexist with superconducting protons and relativistic electrons .
these equations are analogous to the landau model for superfluid helium he . ] , and should also ( after some modifications to incorporate elasticity and possible vortex pinning ) be relevant for the conditions in the inner crust of a mature neutron star . thus we expect the two - stream instability to be generic in dynamical superfluids , possibly limiting the relative flow rates of any multi - fluid system .
our second model problem concerned two fluids confined within an infinitesimally thin spherical shell .
the aim of this model was to assess whether the two - stream instability may be relevant ( perhaps as trigger mechanism ) for pulsar glitches .
the results for this problem demonstrated that the entrainment effect could provide a sufficiently strong coupling for the instability to set in at a relative flow small enough to be astrophysically plausible .
incidentally , the modes that become dynamically unstable in this problem are the superfluid analogues of the inertial r - modes of a rotating single fluid star .
this is interesting since the r - modes are known to be secularly unstable due to the emission of gravitational radiation @xcite .
in fact , the connection between the two instabilities goes even deeper than this since the radiation - driven secular instability is also a variation of the kelvin - helmholtz instability . in that case , the two fluids are the stellar fluid and the radiation .
in order for an instability to be relevant the unstable mode must grow faster than all dissipation timescales . in the case of a superfluid neutron star core
the main dissipation mechanisms are likely to be mutual friction and `` standard '' shear viscosity due to electron - electron scattering . since we have tried to build a plausible case for the two - stream instability to be relevant for pulsar glitches we would like to obtain some rough estimates of the associated dissipation timescales .
to do this we first use an estimate of when mutual friction is likely to dominate the shear viscosity [ due to @xcite ] : @xmath224 where we assume that the wavelength of the mode is @xmath225 .
we can write this as @xmath226 which suggests that the shear viscosity will be the dominant dissipation mechanism for large values of @xmath200 .
for example , for a neutron star rotating with the period of the vela pulsar mutual friction would dominate for @xmath227 or so ( assuming @xmath228 ) . to estimate the shear viscosity damping we can use results obtained for the secular r - mode instability . in particular ,
@xcite have shown that for a uniform density star one has @xmath229 if we use the shear viscosity coefficient for electron - electron scattering @xmath230 we get @xmath231 we want to compare this damping timescale to the growth rate of the unstable modes in our shell toy - model . combining ( [ growth ] ) with ( [ tsv ] )
we estimate that in order to have an instability we must have @xmath232 let us now consider the case of the vela pulsar . estimating the core temperature as @xmath233 k [ roughly two orders of magnitude higher than the observed surface temperature @xcite ] we deduce that only modes with @xmath234 or so are likely to be stabilized by shear viscosity . given that our results indicate that the two - stream instability is active for much smaller values of @xmath200 , cf .
the results shown in figure [ rotdomain ] , we conclude the dissipation is unlikely to suppress the instability in sufficiently young neutron stars
. incidentally , the length scale corresponding to a mode with @xmath235 would be about ten meters .
this is an interesting result since one can show that a large glitch could be explained by a small fraction ( @xmath236 ) of the neutron vortices moving a few tens of metres @xcite .
obviously , the situation changes as the star cools further .
based on the above estimates one can show that shear viscosity will suppress all modes with @xmath201 ( i.e. all unstable modes for the case considered in fig .
[ rotdomain ] ) if the core temperature is below @xmath237 k. this means that the two - stream instability may not be able to overcome the viscous damping in a sufficiently cold neutron star , which is consistent with the absence of glitches in mature pulsars .
we believe that the results of this paper suggest that the superfluid two - stream instability may be relevant in the context of pulsar glitches . if this is , indeed , the case then what is its exact role ?
the answer to this question obviously requires much further work , but it is nevertheless interesting to speculate about some possibilities .
most standard models for glitches are based on the idea of catastrophic vortex unpinning in the inner crust @xcite .
this is an attractive idea since the glitch relaxation ( on a timescale of days to months ) would seem to be well described by vortex creep models @xcite .
an interesting scenario is provided by the thermally induced glitch model discussed by @xcite .
they have shown that the deposit of @xmath238 erg of heat would be sufficient to induce a vela type glitch .
the mechanism that leads to the unpinning of vortices , eg . by the deposit of heat in the crust
, is however not identified .
we believe that the two - stream instability may fill this gap in the current theory .
it should , of course , be pointed out that glitches need not originate in the inner crust .
in particular , @xcite has argued that the vortex pinning is too weak to explain the recurrent vela glitches .
if this argument is correct then the glitches must be due to some mechanism operating in the core fluid . since the model problems we have considered would be relevant for the conditions expected to prevail in the outer core of a mature neutron star ,
our results show that the two - stream instability may serve as a trigger for glitches originating there .
the key requirement for the instability to operate is the presence of a rotational lag .
it is worth pointing out that such a lag will build up both when there is a strong coupling between the two fluids ( i.e. when the vortices are pinned ) and when this coupling is weak .
one would generally expect the strength of this coupling to vary considerably at various depths in the star @xcite , and it is not yet clear to what extent a rotational lag can build up in various regions .
this is , of course , a key issue for future theoretical work on pulsar glitches . one final relevant point concerns the recent observation of a vela size glitch in the anomalous x - ray pulsar 1rxs j170849.0 - 400910 @xcite .
this object has a spin period of 11 s , which means that any feasible glitch model must not rely on the star being rapidly rotating .
what does this mean for our proposal that the two - stream instability may induce a glitch ?
let us assume that the rotational lag builds up at the same rate as the electromagnetic spindown of the main part of the star ( i.e. that the superfluid component does not change its spin rate at all under normal circumstances ) .
then the lag would be @xmath239 after time @xmath27 . if there is a critical value at which a glitch will happen ( corresponding to @xmath240 ) then the interglitch time @xmath241 could be approximated by @xmath242 where @xmath243 is the standard `` pulsar age '' .
this argument implies the following : i ) for @xmath244 we would get @xmath245 .
this ( roughly ) means that only pulsars younger than @xmath194 yr would be seen to glitch during 30 years of observation , which accords well with the fact that only young pulsars are active in this sense .
ii ) there is no restriction on the rotation rate in this scenario ; a star spinning slowly may well exhibit a glitch as long as its spindown rate is fast enough .
this means that one should not be surprised to find glitches in stars with extreme magnetic fields ( magnetars ) .
this paper is only a first probe into what promises to be a rich problem area .
future studies must address issues concerning the effects of different dissipation mechanisms , the nonlinear evolution of the instability , possible experimental verification for superfluid helium etcetera .
these are all very interesting problems which we hope to investigate in the near future .
na and rp acknowledge support from the eu programme improving the human research potential and the socio - economic knowledge base ( research training network contract hprn - ct-2000 - 00137 ) .
na acknowledges support from the leverhulme trust in the form of a prize fellowship , as well as generous hospitality offered by the center for gravitational - wave phenomenology at penn state university .
gc acknowledges partial support from nsf grant phy-0140138 . | this paper provides the first study of a new dynamical instability in superfluids .
this instability is similar to the two - stream instability known to operate in plasmas .
it is analogous to the kelvin - helmholtz instability , but has the distinguishing feature that the two fluids are interpenetrating .
the instability sets in once the relative flow between the two components of the system reaches a critical level .
our analysis is based on the two - fluid equations that have been used to model the dynamics of the outer core of a neutron star , where superfluid neutrons are expected to coexist with superconducting protons and relativistic electrons .
these equations are analogous to the standard landau model for superfluid helium .
we study this instability for two different model problems .
first we analyze a local dispersion relation for waves in a system where one fluid is at rest while the other flows at a constant rate .
this provides a proof of principle of the existence of the two - stream instability for superfluids .
our second model problem concerns two rotating fluids confined within an infinitesimally thin spherical shell .
the aim of this model is to assess whether the two - stream instability may be relevant ( perhaps as a trigger mechanism ) for pulsar glitches .
our results for this problem show that the entrainment effect could provide a sufficiently strong coupling for the instability to set in at a relative flow small enough to be astrophysically plausible . |
analysis of network data is important in a range of disciplines and applications , appearing in such diverse areas as sociology , epidemiology , computer science , and national security , to name a few .
network data here refers to observed edges between nodes , possibly accompanied by additional information on the nodes and/or the edges , for example , edge weights .
one of the fundamental questions in analysis of such data is detecting and modeling community structure within the network .
a lot of algorithmic approaches to community detection have been proposed , particularly in the physics literature ; see @xcite for reviews .
these include various greedy methods such as hierarchical clustering ( see @xcite for a review ) and algorithms based on optimizing a global criterion over all possible partitions , such as normalized cuts @xcite and modularity .
the statistics literature has been more focused on model - based methods , which postulate and fit a probabilistic model for a network with communities .
these include the popular stochastic block model @xcite , its extensions to include varying degree distributions within communities @xcite and overlapping communities , and various latent variable models @xcite .
the stochastic block model is perhaps the most commonly used and best studied model for community detection . for a network with @xmath0 nodes defined by its @xmath1 adjacency matrix @xmath2 , this model postulates that the true node labels @xmath3 are drawn independently from the multinomial distribution with parameter @xmath4 , where @xmath5 for all @xmath6 , and @xmath7 is the number of communities , assumed known .
conditional on the labels , the edge variables @xmath8 for @xmath9 are independent bernoulli variables with @xmath10 = p_{c_i c_j},\ ] ] where @xmath11 $ ] is a @xmath12 symmetric matrix .
the network is undirected , so @xmath13 , and @xmath14 ( no self - loops ) .
the problem of community detection is then to infer the node labels @xmath15 from @xmath2 , which typically also involves estimating @xmath16 and @xmath17 .
there are many extensions of the block model , notably to mixed membership models @xcite , but we will only focus on one extension here that we use later in the paper .
the block model implies the same expected degree for all nodes within a community , which excludes networks with `` hub '' nodes commonly encountered in practice .
the degree - corrected block model @xcite removes this constraint by replacing ( [ bm ] ) with @xmath18 = \theta _ i \theta_j p_{c_i c_j}$ ] , where @xmath19 s are node degree parameters which satisfy an identifiability constraint .
if the degree parameters only take on a discrete number of values , one can think of the degree - corrected block model as a regular block model with a larger number of blocks , but that loses the original interpretation of communities . in @xcite the bernoulli distribution for @xmath8
was replaced by the poisson , primarily for ease of technical derivations , and in fact this is a good approximation for a range of networks @xcite .
fitting block models is nontrivial , especially for large networks , since in principle the problem of optimizing over all possible label assignments is np - hard . in the bayesian framework , markov chain monte carlo methods have been developed , but they only work for networks with a few hundred nodes .
variational methods have also been developed and studied ( see , e.g. , ) , and are generally substantially faster than the gibbs sampling involved in mcmc , but still do not scale to the order of a million nodes .
another bayesian approach based on a belief propagation algorithm was proposed recently by decelle et al .
@xcite , and is comparable to ours in theoretical complexity , but slower in practice ; see more on this in section [ secsimulations].=1 in the non - bayesian framework , a profile likelihood approach was proposed in : since for a given label assignment parameters can be estimated trivially by plug - in , they can be profiled out and the resulting criterion can be maximized over all label assignments by greedy search .
the same method is used in @xcite to fit the degree - corrected block model .
the speed of the profile likelihood algorithms depends on exactly what search method is used and the number of iterations it is run for , but again these generally work well for thousands but not millions of nodes .
a method of moments approach was proposed in , for a large class of network models that includes the block model as a special case .
the generality of this method is an advantage , but it involves counting all occurrences of specific patterns in the graph , which is computationally challenging beyond simple special cases
. some faster approximations for block model fitting based on spectral representations are also available @xcite , but the properties of these approximations are only partially known .
profile likelihood methods have been proven to give consistent estimates of the labels when the degree of the graph grows with the number of nodes , under both the stochastic block models and the degree - corrected version @xcite . to obtain `` strong consistency '' of the labels , that is , the probability of the estimated label vector being equal to the truth converging to 1 , the average graph degree @xmath20 has to grow faster than @xmath21 , where @xmath0 is the number of nodes . to obtain `` weak consistency
, '' that is , the fraction of misclassified nodes converging to 0 , one only needs @xmath22 .
asymptotic behavior of variational methods is studied in @xcite and , and in @xcite this belief propagation method is analyzed for both the sparse [ @xmath23 and the dense ( @xmath24 ) regimes , by nonrigorous cavity methods from physics , and a phase transition threshold , below which the labels can not be recovered , is established . in fact , it is easy to see that consistency is impossible to achieve unless @xmath25 , since otherwise the expected fraction of isolated nodes does not go to 0 . the results one can get for the sparse case , such as @xcite , can only claim that the estimated labels are correlated with the truth better than random guessing , but not that they are consistent . in this paper , for the purposes of theory we focus on consistency and thus necessarily assume that the degree grows with @xmath0 .
however , in practice we find that our methods are very well suited for sparse networks and work well on graphs with quite small degrees .
our main contribution here is a new fast pseudo - likelihood algorithm for fitting the block model , as well as its variation conditional on node degrees that allows for fitting networks with highly variable node degrees within communities .
the idea of pseudo - likelihood dates back to @xcite , and in general amounts to ignoring some of the dependency structure of the data in order to simplify the likelihood and make it more tractable
. the main feature of the adjacency matrix we ignore here is its symmetry ; we also apply block compression , that is , divide the nodes into blocks and only look at the likelihood of the row sums within blocks .
this leads to an accurate and fast approximation to the block model likelihood , which allows us to easily fit block models to networks with tens of millions of nodes .
another major contribution of the paper is the consistency proof of one step of the algorithm .
the proof requires new and somewhat delicate arguments not previously used in consistency proofs for networks ; in particular , we use the device of assuming an initial value that has a certain overlap with the truth , and then show the amount of overlap can be arbitrarily close to purely random . finally , we propose spectral clustering with perturbations , a new clustering method of independent interest which we use to initialize pseudo - likelihood in practice . for sparse networks ,
regular spectral clustering often performs very poorly , likely due to the presence of many disconnected components .
we perturb the network by adding additional weak edges to connect these components , resulting in regularized spectral clustering which performs well under a wide range of settings .
the rest of the paper is organized as follows .
we present the algorithms in section [ secalgorithms ] , and prove asymptotic consistency of pseudo - likelihood in section [ secconsist ] .
the numerical performance of the methods is demonstrated on a range of simulated networks in section [ secsimulations ] and on a network of political blogs in section [ secexample ] .
section [ secdiscuss ] concludes with discussion , and the contains some additional technical results .
the joint likelihood of @xmath2 and @xmath15 could in principle be maximized via the expectation maximization ( em ) algorithm , but the e - step involves optimizing over all possible label assignments , which is np - hard .
instead , we introduce an initial labeling vector @xmath26 , @xmath27 , which partitions the nodes into @xmath7 groups .
note that for convenience we partition into the same number of groups as we assume to exist in the true model , but in principle the same idea can be applied with a different number of groups ; in fact dividing the nodes into @xmath0 groups with a single node in each group instead gives an algorithm equivalent to that of .
the main quantity we work with are the block sums along the columns , @xmath28 for @xmath29 , @xmath30 .
let @xmath31 .
further , let @xmath32 be the @xmath33 matrix with entries @xmath34 given by @xmath35 let @xmath36 be the @xmath37th row of @xmath32 , and let @xmath38 be the @xmath39th column of @xmath17 .
let @xmath40 and @xmath41 .
our approach is based on the following key observations : for each node @xmath6 , conditional on labels @xmath42 with @xmath43 : @xmath44 are mutually independent ; @xmath45 , a sum of independent bernoulli variables , is approximately poisson with mean @xmath46 . with true labels @xmath47 unknown
, each @xmath48 can be viewed as a mixture of poisson vectors , identifiable as long as @xmath49 has no identical rows . by ignoring the dependence among @xmath50 , using the poisson assumption , treating @xmath47 as latent variables , and setting @xmath51
, we can write the pseudo log - likelihood as follows ( up to a constant ) : @xmath52 a pseudo - likelihood estimate of @xmath53 can then be obtained by maximizing @xmath54 .
this can be done via the standard em algorithm for mixture models , which alternates updating parameter values with updating probabilities of node labels .
once the em converges , we update the initial block partition vector @xmath55 to the most likely label for each node as indicated by em , and repeat this process for a fixed number of iterations @xmath56 . for any labeling @xmath55 ,
let @xmath57 , @xmath58 if @xmath59 , @xmath60 and @xmath61 .
we suppress the dependence on @xmath55 whenever there is no ambiguity .
the details of the algorithmic steps can be summarized as follows . _
the pseudo - likelihood algorithm .
_ initialize labels @xmath55 , and let @xmath62 , @xmath63 , @xmath64 , @xmath65 , @xmath66 and @xmath67
. then repeat @xmath56 times : compute the block sums @xmath68 according to ( [ eqbikdef ] ) . using current parameter estimates
@xmath69 and @xmath70 , estimate probabilities for node labels by @xmath71 given label probabilities , update parameter values as follows : @xmath72 return to step 2 unless the parameter estimates have converged .
update labels by @xmath73 and return to step 1 .
update @xmath74 as follows : @xmath75 .
in practice , in step 6 we only include the terms corresponding to @xmath76 greater than some small threshold .
the em method fits a valid mixture model as long as the identifiability condition holds , and is thus guaranteed to converge to a stationary point of the objective function @xcite .
another option is to update labels after every parameter update ( i.e. , skip step 4 ) .
we have found empirically that the algorithm above is more stable , and converges faster . in general
, we only need a few label updates until convergence , and even using @xmath77 ( one - step label update ) gives reasonable results with a good initial value
. the choice of the initial value of @xmath55 , on the other hand , can be important ; see more on this in section [ secinitvalue ] . for networks with hub nodes or those with substantial degree variability within communities ,
the block model can provide a poor fit , essentially dividing the nodes into low - degree and high - degree groups .
this has been both observed empirically @xcite and supported by theory @xcite .
the extension of the block model designed to cope with this situation , the degree - corrected block model @xcite , has an extra degree parameter to be estimated for every node , and writing out a pseudo - likelihood that lends itself to an em - type optimization is more complicated .
however , there is a simple alternative : consider the pseudo - likelihood conditional on the observed node degrees . whether these degrees are similar or not will not then matter , and the fitted parameters will reflect the underlying block structure rather than the similarities in degrees . the conditional pseudo - likelihood is again based on a simple observation : if random variables @xmath78 are independent poisson with means @xmath79 , their distribution conditional on @xmath80 is multinomial . applying this observation to the variables
@xmath81 , we have that their distribution , conditional on labels @xmath15 with @xmath43 and the node degree @xmath82 , is multinomial with parameters @xmath83 , where @xmath84 . the conditional log pseudo - likelihood ( up to a constant )
is then given by @xmath85 and the parameters can be obtained by maximizing this function via the em algorithm for mixture models , as before .
we again repeat the em for a fixed number of iterations , updating the initial partition vector after the em has converged .
the algorithm is then the same as that for unconditional pseudo - likelihood , with steps 2 and 3 replaced by : based on current estimates @xmath86 and @xmath87 , let @xmath88 given label probabilities , update parameter values as follows : @xmath89 we now turn to the question of how to initialize the partition vector @xmath55 .
note that the full likelihood , pseudo - likelihoods @xmath90 and @xmath91 , and other standard objective functions used for community detection such as modularity can all be multi - modal . the numerical results in section [ secsimulations ] suggest that the initial value can not be entirely arbitrary , but the results are not too sensitive to it .
we will quantify this further in section [ secsimulations ] ; here we describe the two options we use as initial values , both of which are of independent interest as clustering algorithms for networks .
one of the simplest possible ways to group nodes in a network is to separate them by degree , say by one - dimensional @xmath7-means clustering applied to the degrees as in @xcite .
this only works for certain types of block models , identifiable from their degree distributions , and in general @xmath7-means does not deal well with data with many ties , which is the case with degrees .
instead , we consider two - dimensional @xmath7-means clustering on the pairs @xmath92 , where @xmath93 is the number of paths of length 2 from node @xmath6 , which can be obtained by summing the rows of @xmath94 . a more sophisticated clustering scheme is based on spectral properties of the adjacency matrix @xmath95 or its graph laplacian .
let @xmath96 be diagonal matrix collecting node degrees .
a common approach is to look at the eigenvectors of the normalized graph laplacian @xmath97 , choosing a small number , say @xmath98 , corresponding to @xmath99 largest ( in absolute value ) eigenvalues , with the largest eigenvalue omitted ; see , for example , @xcite .
these vectors provide an @xmath99-dimensional representation for nodes of the graph , on which we can apply @xmath7-means to find clusters ; this is one of the versions of spectral clustering , which was analyzed in the context of the block model in @xcite .
we found that this version of spectral clustering tends to do poorly at community detection when applied to sparse graphs , say , with expected degree @xmath100 .
the @xmath99-dimensional representation seems to collapse to a few points , likely due to the presence of many disconnected components .
we have found , however , that a simple modification performs surprisingly well , even for values of @xmath101 close to 1 .
the idea is to connect all disconnected components which belong to the same community by adding artificial `` weak '' links . to be precise ,
we `` regularize '' the adjacency matrix @xmath2 by adding @xmath102 multiplied by the adjacency matrix of an erdos
renyi graph on @xmath0 nodes with edge probability @xmath103 , where @xmath104 is a constant .
we found that , empirically , @xmath105 works well for the range of @xmath0 considered in our simulations , and that the results are essentially the same for all @xmath106 thus we make the simplest and computationally cheapest choice of @xmath107 , adding a constant matrix of small values , namely , @xmath108 where @xmath109 is the all - ones @xmath0-vector , to the original adjacency matrix .
the rest of the steps , that is , forming the laplacian , obtaining the spectral representation and applying @xmath7-means , are performed on this regularized version of @xmath2 .
we note that to obtain the spectral representation , one only needs to know how the matrix acts on a given vector ; since @xmath110 , the addition of the constant perturbation does not increase computational complexity .
we will refer to this algorithm as spectral clustering with perturbations ( scp ) , since we perturb the network by adding new , low - weight `` edges . ''
by consistency we mean consistency of node labels ( to be defined precisely below ) under a block model as the size of the graph @xmath0 grows . for the theoretical analysis , we only consider the case of @xmath111 communities .
we condition on the community labels @xmath112 , that is , we treat them as deterministic unknown parameters . for simplicity , here we consider the case of balanced communities , each having @xmath113 nodes .
an extension to the unbalanced case is provided in the supplementary material @xcite .
the assumption of balanced communities naturally leads us to use the class prior estimates @xmath114 in ( [ eqcpl1stepiter ] ) .
we call this assumption ( e ) ( for equal class sizes ) : assume each class contains @xmath113 nodes , and set @xmath114 . without loss of generality
, we can take @xmath115 for @xmath116 . as an intermediate step in proving consistency for the block model introduced in section [ secintro ] ,
we first prove the result for a _ directed _ block model .
recall that for the ( undirected ) block model introduced earlier , one has @xmath117 in the directed case , we assume that all the entries in the adjacency matrix are drawn independently , that is , @xmath118 we will use different symbols for the adjacency and edge - probability matrices in the two cases .
this is to avoid confusion when we need to introduce a coupling between the two models . in both cases ,
we have assumed that diagonal entries of the adjacency matrices are also drawn randomly ( i.e. , we allow for self - loops as valid within - community edges ) .
this is convenient in the analysis with minor effect on the results .
the directed model is a natural extension of the block model when one considers the pseudo - likelihood approach ; in particular , it is the model for which the pseudo - likelihood assumption of independence holds .
it is also a useful model of independent interest in many practical situations , in which there is a natural direction to the link between nodes , for example , in email , web , routing and some social networks .
the model can be traced back to the work of holland and leinhardt @xcite and wang and wong @xcite in which it has been implicitly studied in the context of more general exponential families of distributions for directed random graphs .
our approach is to prove a consistency result for the directed model , with an edge - probability matrix of the form @xmath119 note that the only additional restriction we are imposing is that @xmath120 has the same diagonal entries .
both @xmath121 and @xmath122 depend on @xmath0 and can in principle change with @xmath0 at different rates .
this is a slightly different parametrization from the more conventional @xmath123 , where @xmath124 ( and @xmath16 ) do not depend on @xmath0 , and @xmath125 .
we use this particular parametrization here because we only consider the case @xmath126 , and it makes our results more directly comparable to those obtained in the physics literature , for example , @xcite .
a coupling between the directed and the undirected model that we will introduce allows us to carry the consistency result over to the undirected model , with the edge - probability matrix @xmath127 asymptotically , the two edge - probability matrices have comparable ( to first order ) expected degree and out - in - ratio ( as defined by @xcite ) , under mild assumptions .
the average degrees for @xmath120 and @xmath17 are @xmath128 and @xmath129 , respectively .
the latter is @xmath130 as long as @xmath131 .
the condition is satisfied as soon as the average degree of the directed model has sublinear growth : @xmath132 .
the same holds for out - in - ratios . for our analysis
, we consider an e - step of the cpl algorithm .
it starts from some initial estimates @xmath133 , @xmath134 and @xmath135 of parameters @xmath121 , @xmath122 and @xmath16 , together with an initial labeling @xmath55 , and outputs the label estimates @xmath136,\ ] ] where @xmath137 are the elements of the matrix obtained by row normalization of @xmath138^t$ ] . here
@xmath139 is the confusion matrix as defined in ( [ eqrdef ] ) , and @xmath74 is given by either ( [ eqedgeprobdir ] ) or ( [ eqedgeprobundir ] ) , depending on the model , with @xmath121 and @xmath122 replaced with their estimates @xmath133 and @xmath134 .
the key assumption of our analysis is that the initial labeling has a certain overlap with the truth ( we will show later that the amount of overlap is not important ) . one situation where this might naturally arise is survey data , when some small fraction of nodes has been surveyed about their community membership .
another possibility is to run some other crude algorithm first to obtain a preliminary result .
more formally , we consider an initial labeling @xmath140 , which is balanced ( i.e. , assigns equal number of nodes to each label ) and _ matches exactly @xmath141 labels in community @xmath142 _ , for some @xmath143 .
we do not assume that we know which labels are matched , or the value of @xmath144 .
it is easy to see that this is equivalent to @xmath55 matching exactly @xmath141 labels in each of the two communities .
assuming @xmath141 to be an integer , let @xmath145 denote the collection of such labelings , @xmath146 our goal is to obtain a uniform result guaranteeing the consistency of cpl iteration ( [ eqcpl1stepiter ] ) for any initial labeling in @xmath147 .
in particular , this guarantees consistency for any initial labeling of strength at least @xmath144 , even if it is obtained by an algorithm operating on the same adjacency matrix used by cpl .
as will become clear in the course of the proof of theorem [ thmcpldir ] , although @xmath148 depend on @xmath149 ( which in turn depends on @xmath144 ) and @xmath74 , under the stated ( idealized ) assumptions , we do not need to know their exact values in order to implement rule ( [ eqcpl1stepiter ] ) .
in particular , we do not need to know @xmath144 .
we can plug in any number in @xmath150 for @xmath144 and get the same estimates .
note that the value of @xmath151 corresponds to `` no correlation '' between the true and the initial labeling , whereas @xmath152 and @xmath153 both correspond to perfect correlation ( the labels are either all true or all flipped ) .
let us consider the directed case first . as our measure of performance ( i.e. , the loss function ) , we take the following ( directed - case ) mismatch ratio @xmath154 where @xmath155 are computed based on the directed adjacency matrix @xmath156 , and @xmath157 is the set of permutations of @xmath158 , with @xmath159 accounting for the fact that the labels assigned to the communities are only determined up to a permutation .
the counterpart for the undirected case is denoted by @xmath160 .
note that the notion of consistency based on convergence of this quantity matches the `` weak '' consistency discussed in @xcite , rather than the `` strong '' consistency used by .
define @xmath161 and let @xmath162 , @xmath163 $ ] be the binary entropy function .
let us also consider the collection of estimates @xmath164 which have the same ordering as true parameters @xmath165 , @xmath166 then , we have the following result .
[ thmcpldir ] assume , and let @xmath167 .
let the adjacency matrix @xmath156 be generated according to the directed model ( [ eqdirmod ] ) with edge - probability matrix ( [ eqedgeprobdir ] ) , and assume @xmath168 .
then , there exists a sequence @xmath169 such that @xmath170 and @xmath171 \le\exp \bigl ( { -n\bigl[h ( \gamma)-\kappa_{\gamma}(n)\bigr ] } \bigr),\ ] ] where @xmath172 = o(1)$ ] .
in particular , if @xmath173 , we have @xmath174 and the cpl estimate is uniformly consistent .
we think of @xmath144 as fixed , but it is possible to let @xmath175 , making the problem harder as @xmath176 grows . we still get consistency as long as @xmath177 . in the balanced case
, the cpl iteration has a simple intuitive interpretation , as will become clear during the proof of theorem [ thmcpldir ] .
one starts with an initial assignment of labels to nodes .
then , each node updates its label by taking a majority vote among its neighbors . in the case
where @xmath178 , it is intuitively clear that for @xmath121 large enough , this procedure and only one community is shown . from left to right , we have the initial labeling for a sparse graph @xmath179 , the new labeling for @xmath179 after one cpl iteration , the initial labeling for a dense graph @xmath180 , and the new labeling for @xmath180 after cpl iteration .
nodes with red labels are `` infected , '' that is , their community label is incorrect . for the sparse case ,
cpl iteration spreads the infection , while for the dense case , it has the opposite effect . ]
increases the number of correct labels relative to the initial assignment .
figure [ figmajorvote ] illustrates these ideas . in the general case where @xmath181 , theorem [ thmcpldir ] states that @xmath182 is the key parameter that needs to grow for the procedure to succeed.=1 while the labels are of primary interest in community detection , one may also be interested in consistency of the estimated parameters . under strong consistency in the sense of , consistency of the natural plug - in estimates of the block model parameters follows easily ,
but here we only show weak consistency of the labels . however , in the directed model the pseudo - likelihood function we defined is in fact exactly the likelihood of @xmath183 s .
parameter estimates ( say @xmath133 and @xmath134 ) obtained by the em algorithm converge to a local maximum of this function . as a consequence of theorem [ thmcpldir ] , these estimates are also consistent ( for @xmath121 and @xmath122 ) . since the likelihood is smooth with bounded derivatives , one may be able to use standard arguments to show that the estimated parameters are a unique local maximum in a neighborhood of the truth , and even derive their asymptotic normality along ; see , for example , theorem 6.2.1 , page 384 of .
we do not pursue this direction here .
we now turn to the undirected case .
let @xmath184 [ thmcplundir ] assume , and let @xmath185 .
let the adjacency matrix @xmath2 be generated according to the undirected model ( [ equndirmod ] ) with edge - probability matrix ( [ eqedgeprobundir ] ) , and assume @xmath168 .
in addition , assume @xmath186 for some @xmath187 . then , there exist sequences @xmath188 such that @xmath189 satisfies ( [ equnineq ] ) , with @xmath190 replaced with @xmath191 and @xmath192 satisfies @xmath193 and @xmath194\nonumber\\[-8pt]\\[-8pt ] & & \qquad \le3\exp \bigl ( { -n\bigl[h(\gamma)-\kappa_{\gamma}(n ) \bigr ] } \bigr),\nonumber\end{aligned}\ ] ] where @xmath195 is as defined in theorem [ thmcpldir ] . in particular ,
if @xmath196 , we have @xmath197 , and the cpl estimate is uniformly consistent .
the proofs of both theorems can be found in section [ secproofs ] .
condition ( [ eqagamassump ] ) can be met for a fixed @xmath187 by choosing @xmath144 sufficiently small and an upper bound on @xmath198 in terms of @xmath144 .
for example , for @xmath199 and @xmath200 , we have ( [ eqagamassump ] ) if @xmath201 [ weakerassum ] the parameter @xmath182 controlling consistency is the same as the one reported in @xcite and @xcite .
there the concern is with recovering a labeling which is positively correlated with the truth , and the threshold of success is observed to be @xmath202 .
a similar lower bound was given in for spectral clustering .
here , we are concerned with moving from a positively correlated labeling to one with an asymptotically vanishing mismatch ratio [ i.e. , @xmath203 , which is why we need @xmath173 .
[ remsupparef ] these results can be extended to the case of unbalanced communities .
such an extension is provided for the directed block model in the supplementary material @xcite . there
we consider the model with two communities of sizes @xmath204 and @xmath205 ( not necessarily equal ) and an edge - probability matrix @xmath206 which relaxes our earlier assumption @xmath207 in ( [ eqedgeprobdir ] ) . the class of initial labelings is also enlarged to include those that have @xmath208-overlap with community @xmath37 , that is , @xmath209 , with @xmath210 .
in this situation , one needs more assumptions on the initial estimate @xmath74 used in the cpl iteration than in the balanced case .
supplementary material @xcite gives the details .
while we do not discuss the undirected case in this general setting , ideas used in the proof of theorem [ thmcplundir ] can be used to carry the results from the directed to the undirected case .
here we investigate the performance of both the unconditional and conditional pseudo - likelihood algorithms on simulated networks , as well as that of spectral clustering with perturbations .
we simulate two scenarios , one from the regular stochastic block model and one from the degree - corrected block model , to assess the performance in the presence of hub nodes . throughout this section ,
we fix @xmath211 and @xmath212 .
conditional on the labels , the edges are generated as independent bernoulli variables with probabilities proportional to @xmath213 .
the parameters @xmath214 are drawn independently from the distribution of @xmath215 with @xmath216 , @xmath217 .
we do not enforce the identifiability scaling constraint on @xmath218 at this point as it is absorbed into the scaling of the matrix @xmath17 in ( [ eqpscaling ] ) below .
we consider two values of @xmath219 : @xmath220 , which corresponds to the regular block model , and @xmath221 , which corresponds to a network where 10% of the nodes can be viewed as hubs .
the matrix @xmath17 is constructed as follows .
it is controlled by two parameters : the `` out - in - ratio '' @xmath222 @xcite , which we will vary from 0 to 0.2 , and the weight vector @xmath223 , which determines the relative degrees within communities .
we consider two values of @xmath223 : @xmath224 ( no information about communities is contained in node degrees ) and @xmath225 ( degrees themselves provide relevant information for clustering ) .
if @xmath226 , we set @xmath227 , a diagonal matrix . otherwise , we set the diagonal of @xmath228 to @xmath229 and set all off - diagonal elements to @xmath142 .
we then fix the overall expected network degree @xmath101 , which is the natural parameter to control and which we will vary from 1 to 15 .
then we rescale @xmath228 to obtain this expected degree , giving the final @xmath17 @xmath230 to compare our results to the true labels , we will use normalized mutual information ( nmi ) .
one can think of the confusion matrix @xmath32 as a bivariate probability distribution , and of its row and column sums @xmath231 and @xmath232 as the corresponding marginals .
then the nmi is defined by @xcite as @xmath233 , and is always a number between 0 and 1 ( perfect match ) .
it is useful to have a few benchmark values of nmi for reference : for example , for large @xmath0 , matching @xmath234 , @xmath235 and @xmath236 of the labels correspond to values of nmi of approximately @xmath237 , @xmath238 and @xmath239 , respectively .
all figures show the performance of the following methods : @xmath7-means clustering on 1- and 2-degrees ( dc ) , spectral clustering ( sc ) , spectral clustering with perturbations ( scp ) , unconditional pseudo - likelihood ( upl ) initialized with either dc or scp , and conditional pseudo - likelihood ( cpl ) , with the same two initial values for labelings .
the number of outer iterations for upl and cpl is set to @xmath240 ; @xmath0 , @xmath101 , @xmath219 and the number of replications @xmath241 are specified in the figures . .
] . ] figures [ fignmibeta ] and [ fignmilam ] show results on estimating the node labels with varying @xmath222 and @xmath101 , respectively .
generally , smaller @xmath222 and larger @xmath101 make the problem easier , as we expect . in principle
, degree - based clustering gives no information about the labels with uniform weights @xmath223 , and only a moderate amount of information with nonuniform weights , so it serves as an example of a poor starting value for pseudo - likelihood .
regular spectral clustering performs well with uniform weights , but very poorly with nonuniform weights ; we conjecture that this is due to a limitation of @xmath7-means .
spectral clustering with perturbation , on the other hand , performs very well in all scenarios .
apart from being a useful general method on its own , it also serves as an example of a good starting value for pseudo - likelihood .
figures [ fignmibeta ] and [ fignmilam ] show that pseudo - likelihood achieves large gains over a poor starting value , giving surprisingly good results even when starting from the uninformative degree clustering in the case of @xmath242 .
one exception is unconditional pseudo - likelihood with @xmath243 and @xmath242 , which shows that conditioning is necessary to accommodate variation in degrees when the starting value is not very good .
when spectral clustering with perturbation is used as a starting value , which is already very good , upl and cpl do not have much room to do better , although upl still provides a noticeable improvement , being overall the best method when initialized with scp .
it appears that a good starting value overcomes the limitations of the regular block model for networks with hubs , effectively ruling out the competing solution which divides nodes by degree .
finally , figure [ figtime ] shows run times for all the methods for the case of the regular block model ( @xmath220 ) with different community weights [ @xmath242 and @xmath244 .
the times shown for upl and cpl do not include the time to compute the initial value , which is shown separately . for the case
@xmath242 , all methods take roughly the same amount of time . for the case
@xmath245 , spectral clustering ( sc ) takes considerably more time than the rest . on the other hand
, scp takes nearly the same time as it takes for @xmath224 , and it slightly outperforms dc for larger values of @xmath0 .
this might be explained , in part , by the sparse matrix multiplication required for dc , which is both time and memory - consuming for large @xmath0 .
generally , scp provides an excellent starting value , with low computational complexity in a variety of situations .
we have also done some brief comparisons with the belief propagation ( bp ) method of @xcite .
direct fair comparison is difficult because of the different platform for the belief propagation code and the different way in which it handles initial values ; generally , we found that while the computing time of belief propagation scales with @xmath0 at the same rate as ours , bp is slower by a constant factor of about 10 . in terms of accuracy of community detection , in the examples we tried bp was either similar to or a little worse than pseudo - likelihood .
this dataset on political blogs was compiled by adamic and glance @xcite soon after the 2004 u.s . presidential election .
the nodes are blogs focused on us politics , and the edges are hyperlinks between these blogs .
each blog was manually labeled as liberal or conservative in @xcite , and we treat these as true community labels . following @xcite , we ignore directions of the hyperlinks and analyze the largest connected component of this network , which has 1222 nodes and the average degree of 27 . the distribution of degrees is highly skewed to the right ( the median degree is 13 , and the maximum is 351 ) .
the results in figure [ figblogs ] show that the conditional pseudo - likelihood produces a result closest to the truth , as one would expect in view of highly variable degrees .
its result is also very close to those obtained by profile maximum likelihood for the degree - corrected block model and by two different modularities @xcite .
unconditional pseudo - likelihood , on the other hand , puts high - degree nodes in one group and low - degree nodes in the other .
this is very close to the block model solution @xcite .
this example confirms that the unconditional and conditional pseudo - likelihood methods are correctly fitting the block model and the degree - corrected block model , respectively .
due to symmetry , we can assume without loss of generality that @xmath246 .
similarly , we can assume @xmath247 .
then , for any @xmath248 we have @xmath249 .
these will be our standing assumptions throughout the proofs . to see that the assumptions are not restrictive , one can check that the proof goes through , without change , if @xmath250 and @xmath251 . for the other two cases , namely , @xmath252 and @xmath253 , or @xmath250 and @xmath247
, the proof goes through by switching the estimated labels when matching them with the true labels .
that is , we compare estimated community @xmath142 to true community @xmath254 and vice versa .
these can seen by examining ( [ eqcpl1simp ] ) and the discussion that follows .
let us introduce the following notation : @xmath255 for @xmath256 . as long as @xmath257
, we have @xmath258 under the equal priors assumption ( e ) , the cpl estimate ( [ eqcpl1stepiter ] ) simplifies to @xmath259 where @xmath260 are obtained by block compression of the directed adjacency matrix @xmath156 .
let us focus on @xmath261 from now on . then @xmath262 if @xmath263 for @xmath257 , we have @xmath264 , implying that @xmath265 where @xmath149 is defined in ( [ eqrdef ] ) .
it is then not hard to see that after row normalization of @xmath138^t$ ] , we obtain @xmath266 , and @xmath267 . since by assumption @xmath249 and @xmath252 , it follows that @xmath268 .
then , ( [ eqcpl1simp ] ) is equivalent to @xmath269 .
recalling that @xmath270 , we can write the condition as @xmath271 and @xmath272 . let @xmath273 be the set of all @xmath274 with @xmath257 , that is , @xmath275 for @xmath276 , let @xmath277 be the fraction of mismatches over community @xmath278 .
note that the overall mismatch is @xmath279.\ ] ] since we are focusing on @xmath261 , we are concerned with @xmath280 . in a slight abuse of notation , @xmath281 in ( [ eqoverallmis ] ) is in fact an upper bound on the mismatch ratio as defined in ( [ eqmisddef ] ) , since here we are using a particular permutation the identity .
let us define , for @xmath282 and @xmath283 , @xmath284 then we have @xmath285 where the inequality is due to treating the ambiguous case @xmath286 as error .
we now set out to bound this in probability .
let us start with a tail bound on @xmath287 for fixed @xmath288 and @xmath6 .
[ lemxitailbound ] for any @xmath289 and @xmath290 $ ] , we have @xmath291 \le\exp \biggl ( { -\frac{t^2}{4(a+b ) } } \biggr).\ ] ] we apply the classical bernstein inequality for sums of independent bounded random variables .
let @xmath292 $ ] .
note that @xmath293| \le\max(\alpha _ { ij},1-\alpha_{ij } ) \le1 $ ] . for @xmath261 , we have @xmath294 where @xmath295 is defined based on labeling @xmath55 which correspond to @xmath288 .
in addition , since @xmath296 , we have @xmath297 bernstein inequality implies @xmath298 \le\exp \biggl ( { - \frac{t^2}{2(v+ t/3 ) } } \biggr).\ ] ] noting that for @xmath299 , we have @xmath300 completes the proof . we also need a tail bound on @xmath301 .
let us define @xmath302,\qquad { \bar{p}}_1(r ) = \frac1m\sum_{i=1}^mp_i(r).\ ] ] note that these probabilities do not depend on the particular value of @xmath289 , due to symmetry .
we have the following lemma .
[ lemnisdtailbound ] for @xmath303 , @xmath304
\le\exp \bigl ( { - e m{\bar{p}}_1(r ) u \log u } \bigr).\ ] ] follows from lemma [ lempoitail ] in the , by noting that@xmath305 are independent bernoulli random variables .
now we apply lemma [ lemxitailbound ] with @xmath306 .
note that @xmath307 , for @xmath252 .
noting that the rhs of ( [ eqxitailbound ] ) does not depend on @xmath6 , and using ( [ eqpirdef ] ) , we get @xmath308 the cardinality of the set @xmath309 is @xmath310})^2 $ ] where @xmath311 is the binary entropy function , and @xmath312 is as defined in the statement of the theorem .
( see lemma 6 in the supplementary material @xcite for a proof . )
applying lemma [ lemnisdtailbound ] with @xmath313 and the union bound , we obtain @xmath314\\ & & \qquad \le \exp \bigl\ { { m \bigl [ 2h(\gamma ) -e { \bar{p}}_1(0 ) u_n\log u_n+ 2\kappa _ { \gamma}(n ) \bigr ] } \bigr\}.\end{aligned}\ ] ] pick @xmath315 such that @xmath316 it follows , using @xmath113 , that @xmath317 \le\exp\bigl\ { -\bigl[h(\gamma ) - \kappa_{\gamma}(n)\bigr ] n\bigr\}.\ ] ] by symmetry the same bound holds for @xmath318 .
it follows from ( [ eqoverallmis ] ) that the same holds for @xmath319 .
this completes the proof of theorem [ thmcpldir ] .
recall that @xmath2 and @xmath156 are the adjacency matrices of the undirected and directed cases , respectively .
let us define @xmath320 , @xmath321 , @xmath322 as we did in the directed case , but based on @xmath2 instead of @xmath156 . for example , @xmath323 .
our approach is to introduce a _ deterministic coupling _ between @xmath2 and @xmath156 , which allows us to carry over the results of the directed case .
let @xmath324_{ij } = \cases { 0 , & \quad $ { \widetilde{a}}_{ij } = { \widetilde{a}}_{ji } = 0 $ , \cr 1 , & \quad otherwise.}\ ] ] in other words , the graph of @xmath2 is obtained from that of @xmath156 by removing directions .
note that @xmath325 which matches the relation between ( [ eqedgeprobdir ] ) and ( [ eqedgeprobundir ] ) . from ( [ eqcoupleaat ] )
, we also note that @xmath326 let us now upper - bound @xmath327 in terms of @xmath328 . based on ( [ eqcoupleineq ] ) , only those @xmath329 that are equal to @xmath142
contribute to the upper bound .
more precisely , let @xmath330 , and take @xmath261 from now on .
then @xmath331 we further notice that @xmath332 . to simplify notation ,
let us define @xmath333 where the dependence on @xmath288 is due to @xmath334 being derived from @xmath288 [ recall that @xmath335 .
thus we have shown @xmath336 recall from definition ( [ eqdefagam ] ) that @xmath337 [ lemaistailbound ] fix @xmath338 . for @xmath261 , we have @xmath339 = { \mathbb{p}}\bigl [ { { \widetilde{a}}_{*i}}(\sigma ) > ( 1 + { \varepsilon } ) { a_\gamma}\bigr ] \le
\exp \biggl\ { { -\frac{{\varepsilon}^2}{1+{\varepsilon}/3 } } { a_\gamma}\biggr\}.\ ] ] the equality of the two probabilities follows by symmetry .
let us prove the bound for @xmath340 .
we apply bernstein inequality .
note that @xmath341 = \sum_{j \in{\mathcal{s}}_{11 } } { \mathbb{e}}[{\widetilde{a}}_{ij}]+ \sum _ { j \in{\mathcal{s}}_{12 } } { \mathbb{e}}[{\widetilde{a}}_{ij } ] \\ & = & \sum_{j \in{\mathcal{s}}_{11 } } \frac{a}{m } + \sum _ { j \in{\mathcal{s}}_{12 } } \frac{b}{m } = a\gamma+ b(1-\gamma ) = { a_\gamma}.\end{aligned}\ ] ] since @xmath342 , we obtain @xmath343 \le\exp \biggl ( { -\frac{t^2}{2(\mu+ t/3 ) } } \biggr).\ ] ] setting @xmath344 completes the proof . from ( [ eqxiaxitineq ] )
, it follows that @xmath345 which @xmath346 is the logical or .
this can be seen ( as usual ) by noting that if the rhs does not hold , then @xmath347 , implying @xmath348 . translating to indicator functions , @xmath349 averaging over @xmath261 ( i.e. , applying @xmath350 ) , we get @xmath351 where @xmath352 , and similarly for @xmath353 . note that @xmath354 and @xmath355 , while not independent , have the same distribution by symmetry , so we can focus on bounding one of them .
the key is that each one is a sum of i.i.d .
terms , for example , @xmath356 .
we have a bound on @xmath357 from lemma [ lemnisdtailbound ] .
we can get similar bounds on the @xmath358-terms . to start , let @xmath359,\qquad { \bar{q}}_1(r ) = \frac1m\sum_{i=1}^mq_i(r),\ ] ] similar to ( [ eqpirdef ] ) , and note that these quantities too are independent of the particular choice of @xmath289 .
[ lemqtailbound ] for @xmath303 , @xmath360 \le\exp \bigl ( { - e m{\bar{q}}_1(r ) u \log u } \bigr).\ ] ] follows from lemma [ lempoitail ] in the , by noting that@xmath361 is an independent sequence of bernoulli variables .
the same bound holds for @xmath362 .
recall the definition of @xmath363 from ( [ eqpirdef ] ) . using ( [ eqnqineq ] ) and lemmas [ lemnisdtailbound ] and [ lemqtailbound ]
, we get @xmath364 \biggr ] \\ & & \qquad \le{\mathbb{p}}\biggl [ \sup_{\sigma\in\sigma^\gamma } \frac1 m { \widetilde{n}}_{n,1}(\sigma;r ) \ge e u_n{\bar{p}}_1(r ) \biggr]\\ & & \qquad\quad{}+ 2 { \mathbb{p}}\biggl [ \sup_{\sigma\in\sigma^\gamma } \frac1m{\widetilde{q}}_{n,1 * } ( \sigma;r/2 ) \ge e v_n{\bar{q}}_1(r ) \biggr ] \\ & & \qquad \le \exp \bigl\ { { m \bigl [ 2h(\gamma ) -e { \bar{p}}_1(r ) u_n\log u_n+ 2\kappa _ { \gamma}(n ) \bigr ] } \bigr\}\\ & & \qquad\quad{}+ 2\exp \bigl\ { { m \bigl [ 2h(\gamma ) -e { \bar{q}}_1(r ) v_n\log v_n+ 2\kappa _ { \gamma}(n ) \bigr ] } \bigr\}\end{aligned}\ ] ] as long as @xmath365 . now ,
take @xmath366 , so that lemma [ lemaistailbound ] implies @xmath367 now , in lemma [ lemxitailbound ] , take @xmath368 .
note that the assumption @xmath369 implies @xmath370 .
in addition @xmath371 as before .
thus , the chosen @xmath372 is valid for lemma [ lemxitailbound ]
. furthermore , @xmath373 .
hence , the lemma implies @xmath374 ^ 2\frac { ( a - b)^2}{a+b } } \biggr\}.\ ] ] pick @xmath315 and @xmath375 such that @xmath376 the rest of the argument follows as in the directed case .
this completes the proof of theorem [ thmcplundir ] .
the proposed pseudo - likelihood algorithms provide fast and accurate community detection for a range of settings , including large and sparse networks , contributing to the long history of empirical success of pseudo - likelihood approximations in statistics . for the theoretical analysis ,
we did not focus on the convergence properties of the algorithms , since standard em theory guarantees convergence to a local maximum as long as the underlying poisson or multinomial mixture is identifiable .
the consistency of a single iteration of the algorithm was established for an initial value that is better than purely arbitrary , as long as , roughly speaking , the graph degree grows , and there are two balanced communities with equal expected degrees .
the theory shows that this local maximum is consistent , and unique in a neighborhood of the truth , so in fact there is no need to assume that em has converged to the global maximum , an assumption which is usually made in analyzing em - based estimates . the theoretical analysis can be extended to the general two - community model with possibly unbalanced communities , as detailed in the supplementary material @xcite .
extending our argument to more than two communities also seems possible , but that would require extremely meticulous tracking of a large number of terms which we did not pursue .
we conjecture that additional results may be obtained under weaker assumptions if one focuses simply on estimating the parameters of the block model rather than consistency of the labels , just like one can obtain results for a labeling correlated with the truth ( instead of consistent ) under weaker assumptions discussed in remark [ weakerassum ] .
for example , in a very recent paper @xcite , results are obtained under very weak assumptions for the mean squared error of estimating the block model parameter matrix @xmath17 ( which in itself does not guarantee consistency of the labels ) .
while the primary interest in community detection is estimating the labels rather than the parameters , we plan to investigate this further to see if and how our conditions can be relaxed .
while in theory any `` reasonable '' initial value guarantees convergence , in practice the choice of initial value is still important , and we have investigated a number of options empirically .
spectral clustering with perturbations , which we introduced primarily as a method to initialize pseudo - likelihood , deserves more study , both empirically ( e.g. , investigating the optimal choice of the tuning parameter ) , and theoretically .
this is also a topic for future work .
here is a lemma which we used quite often in proving consistency results in section [ secproofs ] .
[ lempoitail ] consider @xmath377 to be independent bernoulli variables with @xmath378 = p_i$ ] .
let @xmath379 , @xmath380 = \sum_{i=1}^mp_i$ ] and @xmath381 .
then , for any @xmath303 , we have @xmath382 we apply a direct chernoff bound .
let @xmath383 . then , by a result of hoeffding @xcite ( also see @xcite ) , @xmath384 for any convex function @xmath385 . letting @xmath386 , we obtain for @xmath387 , @xmath388 where we have used @xmath389 .
the rhs is the chernoff bound for a poisson random variable with mean @xmath390 , and can be optimized to yield @xmath391 letting @xmath392 for @xmath303 and noting that @xmath393 , we get @xmath394 which is the desired bound .
we would like to thank roman vershynin ( mathematics , university of michigan ) for highly illuminating discussions . | many algorithms have been proposed for fitting network models with communities , but most of them do not scale well to large networks , and often fail on sparse networks . here
we propose a new fast pseudo - likelihood method for fitting the stochastic block model for networks , as well as a variant that allows for an arbitrary degree distribution by conditioning on degrees .
we show that the algorithms perform well under a range of settings , including on very sparse networks , and illustrate on the example of a network of political blogs .
we also propose spectral clustering with perturbations , a method of independent interest , which works well on sparse networks where regular spectral clustering fails , and use it to provide an initial value for pseudo - likelihood .
we prove that pseudo - likelihood provides consistent estimates of the communities under a mild condition on the starting value , for the case of a block model with two communities . , , + |
knowledge of earth s internal structure and composition has primarily been obtained through complimentary approaches of geophysical and geochemical investigation .
however , debate still persists about the mechanisms responsible for generating the chemical and physical characteristics of earth s mantle observed today , as well as their evolution through time .
the large - scale composition of the mantle may be inferred in part from mantle - derived melts erupted at or near the earth s surface , including mid - ocean ridge basalts ( morb ) and ocean island basalts ( oib ) .
therefore , analyses of oceanic basalts remain a key component in understanding the geochemical complexities of the mantle .
trace element ratios , radiogenic isotope variations ( e.g. , sr , nd , hf , os , pb ) , and noble gas systematics have traditionally been studied in mantle - derived samples to better understand the chemical composition and physical processes occurring within the mantle throughout earth history ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
ocean island basalts exhibit systematic differences in isotope compositions when compared with morb ; this has generally been attributed to mantle heterogeneities arising from the recycling of various materials into oib mantle source regions ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
various different mantle endmembers have been proposed to explain the variable sr , nd , and pb isotope ratios , in that each plume samples chemically distinct mantle sources @xcite ; this has been recently extended to hf ( e.g. , * ? ? ?
* ) and os isotopes ( e.g. , * ? ? ?
the main compositional mantle endmembers include dmm ( depleted morb mantle ) , em-1 ( enriched mantle 1 ) , em-2 ( enriched mantle 2 ) , and himu ( `` high-@xmath4 '' , where @xmath4 is defined as the time - integrated ^238^u/^204^pb ) .
each endmember is defined in terms of their relative radiogenic sr and pb isotope characteristics : dmm displays low ^87^sr/^86^sr and relatively unradiogenic pb , em-1 has intermediate ^87^sr/^86^sr and low ^206^pb/^204^pb , em-2 shows high ^87^sr/^86^sr and intermediate ^206^pb/^204^pb , and himu is defined by low ^87^sr/^86^sr and high ^206^pb/^204^pb @xcite .
these geochemical characteristics are thought to be a reflection of specific lithologies present as a component of the source : em-1 may incorporate recycled lower continental crust material or pelagic sediments @xcite or delaminated subcontinental lithosphere @xcite , em-2 may arise from the recycling of continental - derived sediments @xcite , and himu may involve recycled oceanic crust and lithospheric mantle @xcite .
it has even been proposed that an ancient marine carbonate component may exist in the himu mantle source @xcite .
the traditional mantle endmembers represent extremes in terms of isotopic composition , with each oib location potentially representing a unique mixture of mantle components .
however , linking observations of chemical heterogeneity of oib with the mechanisms responsible for generating such variations is currently a major challenge in mantle geochemistry .
in contrast to radiogenic isotope systems , the mantle end - members are less well defined by stable isotope variations ; although o , mg , and ca show possible recycled signatures ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , generally it is assumed that the mantle is homogeneous with respect to most stable isotope systems . to date , this has been the case for silicon isotopes .
for example , samples of and from the shallow mantle ( peridotites and morb ) are assumed to be representative of the si isotope composition of bulk silicate earth ( bse ) , with any si isotope heterogeneity being obliterated through convective mantle mixing @xcite .
however , a systematic study of the si isotope composition of oib has not previously been performed ; only a limited number of high - precision analyses are present in the literature ( see * ? ? ?
* for a recent review ) .
given the implications of possible mantle heterogeneities on a wide variety of mantle processes and the importance of an accurate characterization of the isotopic composition of bse for geo- and cosmo - chemical comparison purposes , a dedicated investigation of potential si isotope variations in oib is warranted .
the si isotope system is a potentially useful tool with which to investigate recycling of terrestrial surface materials within the mantle .
silicon is a key element in silicate minerals and is composed of three stable isotopes : ^28^si ( 92.23% ) , ^29^si ( 4.68% ) , and ^30^si ( 3.09% ) .
relatively large si isotope fractionations have been observed as the result of low - temperature bio- and geo - chemical surficial / supergene processes , with the light si isotopes becoming preferentially enriched in precipitated silica phases compared to dissolved si in the fluid phase @xcite .
previous studies of the si isotope systematics in low - temperature environments , including chert formation @xcite , diatom formation @xcite , and silicate weathering @xcite , have shown that si isotope variations can range over several per mil in surface environments .
in contrast , high - temperature igneous processes typically result in relatively small fractionations ; terrestrial igneous materials display only a limited range ( @xmath00.2 per mil on the ^30^si/^28^si ratio ) of si isotope compositions @xcite , with the bulk silicate earth being relatively homogeneous ( @xmath1 ^ 30^si = @xmath20.29 @xmath3 0.07 , 2 sd ; * ? ? ?
partial mantle melting to form basaltic melts does not appear to generate large si isotope fractionations @xcite compared to low - temperature processes , so incorporation of recycled surface materials in the mantle could produce si isotope variations that may be observed in oib . here
we present high - precision si isotope data obtained by multi - collector inductively - coupled - plasma mass - spectrometry ( mc - icp - ms ) for a suite of oib , including samples representative of the em-1 , em-2 , and himu mantle endmembers , to quantify potential si isotope variations in the mantle and assess the processes that may generate si isotope variability .
a set of 46 oib representing locations in the pacific , atlantic , and indian oceans were selected for si isotope analysis . in particular ,
samples were selected to represent the proposed mantle endmembers , including eight samples from pitcairn ( em-1 ) , eleven samples from samoa ( em-2 ) , and four samples from mangaia ( himu ) .
additional samples represent geographic locations with unique geochemical features , such as primitive noble gas compositions , or areas that display less extreme mantle endmember compositions .
for example , three samples were selected from so nicolau , one of the northern islands of cape verde that has been identified as himu - like based on radiogenic pb isotope compositions and low ^87^sr/^86^sr , but that also exhibits high ^3^he/^4^he ( @xmath5 r / r@xmath6 ) , possibly indicative of a primitive lower mantle component @xcite .
similarly , two samples were selected from st .
helena , which has been identified as himu - like based on radiogenic isotope compositions but displays low ^3^he/^4^he @xcite .
six samples were selected from the canary islands , particularly the islands of el hierro and la palma , whose mantle sources have been interpreted to contain distinct proportions of recycled oceanic crust and lithospheric mantle based on he o sr
nd os pb isotope systematics @xcite .
four icelandic samples from volcanic complexes in the northern rift zone ( nrz ) that display depleted tholeiitic compositions were selected .
iceland is one of the most well - studied oib - type localities due to its association with both a spreading center ( the mid - atlantic ridge ) and a deep - seated mantle plume @xcite . the relatively unradiogenic he and ne isotope ratios measured previously in these samples compared to global oib is possible evidence for a deep , undegassed mantle source @xcite . ^29^si versus @xmath1 ^ 30^si for oib .
all data are within error ( @xmath3 2 se ) of the calculated mass - dependent equilibrium ( solid line , slope 0.5178 ) and kinetic ( dashed line , slope 0.5092 ) fractionation lines . ]
lllrrrrrrrr bhvo-2 & hawaii & pacific & @xmath20.15 & 0.07 & 0.01 & @xmath20.30 & 0.12 & 0.01 & 76 & 49.90 + ahanemo2d20-b & galapagos & pacific & @xmath20.14 & 0.06 & 0.02 & @xmath20.30 & 0.13 & 0.05 & 6 & 48.89 + isi09 - 9 & iceland & atlantic & @xmath20.22 & 0.08 & 0.03 & @xmath20.41 & 0.15 & 0.06 & 6 & 44.34 + kbd408702 & iceland - kambsfell & atlantic & @xmath20.23 & 0.14 & 0.05 & @xmath20.44 & 0.19 & 0.07 & 7 & 47.76 + kbd408709 & iceland - kistufell & atlantic & @xmath20.20 & 0.09 & 0.04 & @xmath20.40 & 0.13 & 0.06 & 5 & 48.26 + kbd408729 & iceland - blafjall & atlantic & @xmath20.17 & 0.08 & 0.04 & @xmath20.36 & 0.15 & 0.08 & 4 & 51.19 + _ * average - iceland * _ & & & * @xmath20.21 * & * 0.06 * & * 0.03 * & * @xmath20.40 * & * 0.06 * & * 0.03 * & * 4 * & - + mg-1001 # 1 & mangaia & pacific & @xmath20.23 & 0.08 & 0.02 & @xmath20.46 & 0.16 & 0.05 & 10 & 43.57 + mg-1001 # 2 & mangaia & pacific & @xmath20.20 & 0.07 & 0.02 & @xmath20.38 & 0.14 & 0.04 & 10 & 43.57 + _ average - mg-1001 _ & & & @xmath20.21 & 0.04 & 0.03 & @xmath20.42 & 0.11 & 0.08 & 2 & 43.57 + mg-1002 # 1 & mangaia & pacific & @xmath20.20 & 0.05 & 0.02 & @xmath20.39 & 0.12 & 0.04 & 10 & 43.35 + mg-1002 # 2 & mangaia & pacific & @xmath20.19 & 0.12 & 0.05 & @xmath20.36 & 0.17 & 0.08 & 5 & 43.35 + _ average - mg-1002 _ & & & @xmath20.20 & 0.02 & 0.02 & @xmath20.37 & 0.05 & 0.04 & 2 & 43.35 + mg-1006 # 1 & mangaia & pacific & @xmath20.15 & 0.04 & 0.02 & @xmath20.30 & 0.09 & 0.04 & 6 & 43.23 + mg-1006 # 2 & mangaia & pacific & @xmath20.14 & 0.04 & 0.02 & @xmath20.26 & 0.07 & 0.03 & 6 & 43.23 + _ average - mg-1006 _ & & & @xmath20.15 & 0.01 & 0.01 & @xmath20.28 & 0.05 & 0.04 & 2 & 43.23 + mg-1008 & mangaia & pacific & @xmath20.20 & 0.05 & 0.02 & @xmath20.38 & 0.07 & 0.03 & 6 & 43.00 + _ * average - mangaia * _ & & & * @xmath20.19 * & * 0.06 * & * 0.03 * & * @xmath20.36 * & * 0.12 * & * 0.06 * & * 4 * & - + fe0903 & canary - fuerteventura & atlantic & @xmath20.12 & 0.04 & 0.02 & @xmath20.24 & 0.05 & 0.02 & 6 & 47.05 + eh03 & canary - el hierro & atlantic & @xmath20.13 & 0.05 & 0.02 & @xmath20.24 & 0.10 & 0.04 & 6 & 42.74 + eh12 & canary - el hierro & atlantic & @xmath20.14 & 0.01 & 0.01 & @xmath20.24 & 0.06 & 0.03 & 6 & 42.08 + eh15 & canary - el hierro & atlantic & @xmath20.17 & 0.02 & 0.01 & @xmath20.34 & 0.11 & 0.05 & 6 & 42.27 + _ * average - el hierro * _ & & & * @xmath20.15 * & * 0.04 * & * 0.02 * & * @xmath20.27 * & * 0.12 * & * 0.07 * & * 3 * & - + lp03 & canary - la palma & atlantic & @xmath20.16 & 0.05 & 0.02 & @xmath20.36 & 0.05 & 0.02 & 6 & 39.04 + lp15 & canary - la palma & atlantic & @xmath20.16 & 0.05 & 0.02 & @xmath20.36 & 0.11 & 0.04 & 6 & 43.23 + _ * average - la palma * _ & & & * @xmath20.16 * & * 0.00 * & * 0.00 * & * @xmath20.36 * & * 0.00 * & * 0.00 * & * 2 * & - + cv sn 98 - 01 & cape verde - nicolau & atlantic & @xmath20.21 & 0.09 & 0.04 & @xmath20.40 & 0.19 & 0.09 & 4 & 41.01 + cv sn 98 - 15 & cape verde - nicolau & atlantic & @xmath20.17 & 0.06 & 0.03 & @xmath20.36 & 0.08 & 0.04 & 4 & 39.50 + cv sn 98 - 18 & cape verde - nicolau & atlantic & @xmath20.22 & 0.08 & 0.04 & @xmath20.40 & 0.15 & 0.09 & 3 & 44.30 + _ * average - cape verde * _ & & & * @xmath20.20 * & * 0.05 * & * 0.03 * & * @xmath20.39 * & * 0.04 * & * 0.02 * & * 3 * & - + sh25 & st helena & atlantic & @xmath20.15 & 0.12 & 0.05 & @xmath20.29 & 0.10 & 0.05 & 5 & 44.95 + bm.1911,1626(1 ) & st helena & atlantic & @xmath20.15 & 0.04 & 0.02 & @xmath20.30 & 0.13 & 0.06 & 5 & 45.56 + _ * average - st helena * _ & & & @xmath20.15 & 0.01 & 0.01 & @xmath20.30 & 0.01 & 0.01 & 2 & - + pit 1 & pitcairn & pacific & @xmath20.13 & 0.04 & 0.02 & @xmath20.26 & 0.10 & 0.04 & 6 & 48.10 + pit 3 & pitcairn & pacific & @xmath20.14 & 0.07 & 0.03 & @xmath20.29 & 0.08 & 0.04 & 5 & 48.73 + pit 4a & pitcairn & pacific & @xmath20.12 & 0.06 & 0.02 & @xmath20.25 & 0.08 & 0.03 & 6 & 48.91 + pit 6 & pitcairn & pacific & @xmath20.17 & 0.06 & 0.03 & @xmath20.31 & 0.08 & 0.03 & 6 & 49.54 + pit 7 & pitcairn & pacific & @xmath20.15 & 0.12 & 0.05 & @xmath20.30 & 0.20 & 0.08 & 6 & 48.15 + pit 8 & pitcairn & pacific & @xmath20.15 & 0.08 & 0.03 & @xmath20.31 & 0.13 & 0.05 & 6 & 46.32 + pit 12 & pitcairn & pacific & @xmath20.16 & 0.10 & 0.04 & @xmath20.30 & 0.22 & 0.09 & 6 & 49.22 + pit 16 & pitcairn & pacific & @xmath20.11 & 0.06 & 0.03 & @xmath20.22 & 0.04 & 0.02 & 6 & 48.71 + _ * average - pitcairn * _ & & & * @xmath20.14 * & * 0.04 * & * 0.01 * & * @xmath20.28 * & * 0.07 * & * 0.02 * & * 8 * & - + ag132 & kerguelen & indian & @xmath20.19 & 0.06 & 0.04 & @xmath20.31 & 0.10 & 0.06 & 3 & 45.08 + bm1962 128 ( 114 ) & tristan da cunha & atlantic & @xmath20.17 & 0.09 & 0.04 & @xmath20.31 & 0.19 & 0.07 & 7 & 43.18 + alr 40 g & gough & atlantic & @xmath20.21 & 0.09 & 0.04 & @xmath20.35 & 0.17 & 0.08 & 4 & 48.54 + bv2 & bouvet & atlantic & @xmath20.17 & 0.07 & 0.03 & @xmath20.35 & 0.11 & 0.05 & 6 & 55.16 + ofu-04 - 14 & samoa - ofu & pacific & @xmath20.17 & 0.08 & 0.03 & @xmath20.31 & 0.10 & 0.03 & 9 & 45.36 + alia-115 - 03 & samoa - savaii & pacific & @xmath20.15 & 0.12 & 0.05 & @xmath20.26 & 0.19 & 0.08 & 6 & 50.31 + alia-115 - 18 & samoa - savaii & pacific & @xmath20.16 & 0.06 & 0.02 & @xmath20.28 & 0.06 & 0.02 & 6 & 54.87 + t16 & samoa - tau & pacific & @xmath20.18 & 0.09 & 0.04 & @xmath20.34 & 0.10 & 0.04 & 6 & 46.71 + t21 & samoa - tau & pacific & @xmath20.16 & 0.08 & 0.03 & @xmath20.32 & 0.19 & 0.08 & 6 & 48.61 + t38 & samoa - tau & pacific & @xmath20.17 & 0.03 & 0.01 & @xmath20.32 & 0.06 & 0.03 & 5 & 45.34 + 74 - 4 & samoa - tau & pacific & @xmath20.18 & 0.04 & 0.02 & @xmath20.33 & 0.05 & 0.02 & 5 & 45.52 + 63 - 2 & samoa - vailuluu & pacific & @xmath20.19 & 0.04 & 0.02 & @xmath20.34 & 0.08 & 0.03 & 6 & 46.14 + 76 - 9 & samoa - malumalu & pacific & @xmath20.17 & 0.05 & 0.02 & @xmath20.36 & 0.07 & 0.03 & 6 & 45.91 + 78 - 1 & samoa - malumalu & pacific & @xmath20.15 & 0.07 & 0.03 & @xmath20.28 & 0.05 & 0.02 & 6 & 45.54 + 128 - 21 & samoa - taumatau & pacific & @xmath20.12 & 0.05 & 0.02 & @xmath20.27 & 0.07 & 0.03 & 6 & 49.82 + _ *
average - samoa * _ & & & * @xmath20.16 * & * 0.04 * & * 0.01 * & * @xmath20.31 * & * 0.07 * & * 0.02 * & * 11 * & - + aco95 - 3 ( 1562 ad ) & azores - so miguel & atlantic & @xmath20.17 & 0.06 & 0.03 & @xmath20.33 & 0.15 & 0.07 & 5 & 47.81 + aco2000 - 51 & azores - so jorge & atlantic & @xmath20.14 & 0.15 & 0.08 & @xmath20.28 & 0.21 & 0.10 & 4 & 45.21 + _ * average - azores * _ & & & * @xmath20.16 * & * 0.04 * & * 0.02 * & * @xmath20.30 * & * 0.07 * & * 0.05 * & * 2 * & - + eruption 1931 & la reunion & indian & @xmath20.14 & 0.14 & 0.06 & @xmath20.30 & 0.12 & 0.05 & 5 & 43.91 + * mineral separates * & & & & & & & & & & + mg-1001 & zeolite & & @xmath20.07 & 0.07 & 0.04 & 0.00 & 0.15 & 0.08 & 4 & 7.40 + mg-1001 & pyroxene & & @xmath20.22 & 0.10 & 0.04 & @xmath20.42 & 0.21 & 0.09 & 5 & 52.70 + when possible , unaltered samples ( based on visual inspection and absence of secondary minerals ) were chosen for analysis in order to minimize the potential effects of post - eruptive weathering , which is a process that could alter the si isotope composition of basalts . in many cases samples were fresh submarine or subglacial basaltic glasses .
major- and trace - element abundances and radiogenic isotopic ratios for the present sample set have been previously reported elsewhere @xcite .
the alkali fusion sample dissolution and chemical purification methods used here have been described recently @xcite , and are based on the methods first described by @xcite .
samples were ground to a fine powder using an agate mortar ; at least 1 g of each whole rock sample was processed to ensure representative sampling .
samples were then digested by an alkali fusion technique .
this sample digestion method avoids the use of hf , which can lead to the loss of si in the form of volatile sif complexes following conventional acid digestion procedures .
approximately 10 mg of powdered sample was combined with @xmath0200 mg naoh flux in an ag crucible and heated for 12 minutes in a muffle furnace at 720 c. the resulting fused sample was dissolved using milli - q h@xmath7o and weakly acidified to 1% hno@xmath8 to obtain a si stock solution sufficiently dilute to avoid silicic acid polymerization .
the samples were then purified for si isotope analysis through ion - exchange chromatography using 1.8 ml biorad ag50 x12 ( 200 - 400 mesh ) cation exchange resin loaded in polyprep columns .
although the speciation of dissolved si in water is ph dependent @xcite , it occurs as neutral or anionic species at low to neutral phs and thus is not retained by the resin whereas cationic matrix elements are efficiently separated . quantitative recovery of si was achieved through elution by 5 ml milli - q h@xmath7o .
column loads were calculated to yield final solution si concentrations of 2 ppm for isotope analyses . 2 se ) .
the dashed line and shaded box represent the estimate ( @xmath3 2 sd ) for the si isotope composition of the bulk silicate earth ( bse ) as given by @xcite .
most samples fall in the range of the estimate for bse , but some himu and iceland localities exhibit systematically lower @xmath1 ^ 30^si values . ]
silicon isotope ratios were measured on thermo scientific neptune plus mc - icp - ms instruments at washington university in st .
louis and at the institut de physique du globe de paris .
each measurement represents 30 cycles with an 8.389 s integration time .
sample solutions were introduced using a wet plasma system and measurements were made in medium resolution on peak shoulder to avoid the large ^14^n^16^o interference on mass 30 .
sample bracketing was used to correct for instrumental mass bias ; the quartz sand standard nbs28 ( nist rm8546 ) and the well - characterized basalt geostandard bhvo-2 were each used as bracketing standards to obtain si isotope measurements during separate analytical sessions and were subjected to the same dissolution and si purification procedure as the samples .
silicon isotope compositions are expressed relative to a bracketing standard ( nbs28 or bhvo-2 ) using the following delta notation : @xmath9 where x = 29 or 30 . reported isotope ratios are averages of repeated measurements of each sample ( `` n '' in table 1 ) , and errors are determined from these repeated measurements . in a @xmath1 ^ 29^si versus @xmath1 ^ 30^si plot ( figure 1 ) , all of the samples measured in this study fall within error of the calculated mass - dependent equilibrium ( slope 0.5178 ) and kinetic ( slope 0.5092 ) fractionation lines , and no si isotope anomalies have been observed in terrestrial samples @xcite .
therefore , the remainder of the text discusses the si isotope data in terms of ^30^si .
silicon isotope data relative to nbs28 as the bracketing standard are reported in table 1 and figure 2 .
the errors discussed throughout the text are 2 standard error ( 2 se ) unless otherwise specified . to assess the reproducibility of the data , the si isotope composition of the well - characterized basalt geostandard bhvo-2 was measured during each analytical session .
the bhvo-2 value measured here ( @xmath1 ^ 30^si = @xmath20.30 @xmath3 0.01 ) represents 76 measurements during 13 separate analytical sessions and is in good agreement with previously published literature data ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
on average , @xmath1 ^ 30^si values for oib ( @xmath20.32 @xmath3 0.09 , 2 sd ) , including the em-1 endmember pitcairn ( @xmath20.28 @xmath3 0.07 , 2 sd ) and the em-2 endmember samoa ( @xmath20.31 @xmath3 0.07 , 2 sd ) , are in general agreement with previous estimates for the @xmath1 ^ 30^si value of bse ( @xmath1 ^ 30^si = @xmath20.29 @xmath3 0.07 , 2 sd ; * ? ? ? * ) .
in addition , no variation in si isotope composition with ocean basin is observable .
however , some locations exhibit light si isotope enrichments relative to morb ; on average , oib from iceland nrz ( @xmath1 ^ 30^si = @xmath20.40 @xmath3 0.06 , 2 sd ) and the himu localities mangaia ( cook islands ; @xmath1 ^ 30^si = @xmath20.36 @xmath3 0.12 , 2 sd ) , la palma ( canary islands ; @xmath1 ^ 30^si = @xmath20.36 @xmath3 0.00 , 2 sd ) , and cape verde ( @xmath1 ^ 30^si = @xmath20.39 @xmath3 0.04 , 2 sd ) are lighter than the other oib measured in this study ( on average , @xmath1 ^ 30^si = @xmath20.30 @xmath3 0.06 , 2 sd ) .
there are two notable exceptions within these localities with low @xmath1 ^ 30^si .
one sample from mangaia ( mg-1006 ) has anomalously high @xmath1 ^ 30^si , but of the mangaia group it has the lowest cao / al@xmath7o@xmath8 value .
additionally , one el hierro sample ( eh15 ) has high @xmath1 ^ 30^si relative to the other el hierro samples ; this sample has the lowest @xmath1 ^ 18^o of the group .
llrrrrrrr nbs28 & & 0.12 & 0.09 & 0.04 & 0.24 & 0.13 & 0.05 & 6 + isi09 - 9 & iceland & @xmath20.05 & 0.05 & 0.02 & @xmath20.10 & 0.10 & 0.04 & 6 + kbd408702 & iceland & @xmath20.03 & 0.07 & 0.03 & @xmath20.09 & 0.12 & 0.05 & 6 + _ * average - iceland * _ & & * @xmath20.04 * & * 0.02 * & - & * @xmath20.10 * & * 0.01 * & - & * 2 * + mg-1001 # 1 & mangaia & @xmath20.04 & 0.10 & 0.04 & @xmath20.07 & 0.21 & 0.08 & 6 + mg-1001 # 2 & mangaia & @xmath20.05 & 0.14 & 0.05 & @xmath20.10 & 0.20 & 0.07 & 6 + _ average - mg-1001 _ & & @xmath20.05 & 0.01 & 0.01 & @xmath20.08 & 0.04 & 0.03 & 2 + mg-1002 # 1 & mangaia & @xmath20.02 & 0.06 & 0.02 & @xmath20.01 & 0.08 & 0.03 & 6 + _ * average - mangaia * _ & & * @xmath20.03 * & * 0.03 * & - & * @xmath20.05 * & * 0.10 * & - & * 2 * + sh25 & st helena & 0.02 & 0.12 & 0.05 & @xmath20.01 & 0.16 & 0.06 & 6 + pit 1 & pitcairn & 0.01 & 0.10 & 0.04 & 0.00 & 0.10 & 0.04 & 6 + ofu-04 - 14 & samoa - ofu & @xmath20.02 & 0.07 & 0.03 & @xmath20.03 & 0.12 & 0.05 & 6 + aco2000 - 51 & azores - so jorge & @xmath20.01 & 0.06 & 0.02 & @xmath20.01 & 0.21 & 0.08 & 6 + 2 se ) . the dashed line and shaded box represent the si isotope composition of bhvo-2 ( @xmath3 2 se ) , which is used as a proxy for bse .
these additional data directly compare selected oib samples to further distinguish si isotope compositions and confirm the variations observed in figure 2 . ] in order to directly measure the variations in si isotope composition among oib , the basalt geostandard bhvo-2 was used as the bracketing standard in a separate measurement session ; these data are denoted @xmath1 ^ 30^si@xmath10 and are presented in table 2 and figure 3 .
bhvo-2 was chosen as a bracketing standard because its si isotope composition is well characterized in the literature and it has previously been found to have the same si isotope composition as the estimated value for bse @xcite .
bhvo-2 is an oib from kilauea volcano , taken from a surface layer of the pahoehoe lava that overflowed from the halemaumau crater in fall 1919 and has 7.23 wt.% mgo .
this direct comparison between oib makes non - zero isotopic offsets easier to resolve since it eliminates a source of analytical error ; when comparing data bracketed using the traditional nbs28 international si isotope standard , the propagation of error compounds the uncertainty . when measured against bhvo-2 , samples from iceland ( @xmath1 ^ 30^si@xmath10 = @xmath20.10 @xmath3 0.01 ) and mangaia ( @xmath1 ^ 30^si@xmath10 = @xmath20.05 @xmath3 0.10 ) are relatively lighter in si isotopes , while samples from st .
helena , pitcairn , samoa , and the azores are indistinguishable from bhvo-2 ( on average , @xmath1 ^ 30^si@xmath10 = @xmath20.01 @xmath3 0.02 ) .
these results are consistent with the data obtained through bracketing with the traditional si isotope standard nbs28 and confirm the observed si isotope variations in oib .
at first order , oib exhibit generally uniform si isotope compositions that are in good agreement with previous estimates for the @xmath1 ^ 30^si value of bse . however , small variations in the data are present , with some localities enriched in light si isotopes by @xmath00.1 on @xmath1 ^ 30^si compared to the estimate for bse .
although this difference is small , it could have significant implications for mantle processes responsible for creating and preserving such heterogeneity .
below , we discuss the possible causes of the si isotope variability observed in oib , including post - eruptive alteration , magmatic differentiation , sampling of a primitive mantle reservoir , crustal assimilation , and recycling of crust and lithospheric mantle . where possible , unaltered samples were chosen to minimize the effects of post - eruptive alteration . however , mangaia volcanic rocks have crystallization ages of @xmath020 myrs and therefore have undergone surface weathering processes @xcite .
although minor chemical weathering should not significantly affect the si isotope composition of the bulk rock @xcite , @xcite found that basalt weathering may result in the formation of secondary minerals that are isotopically light relative to the dissolved si in the fluid phase . to assess the possible contribution of isotopically fractionated secondary minerals on the bulk si isotope composition of mangaia samples , zeolite and pyroxene mineral separates from one mangaia sample ( mg-1001 )
were selected for si isotope analysis by mc - icp - ms .
the si isotope compositions of the mg-1001 mineral separates are reported in table 1 .
the pyroxene fraction ( @xmath1 ^ 30^si = @xmath20.42 @xmath3 0.09 ) is indistinguishable from the mg-1001 whole - rock ( @xmath1 ^ 30^si = @xmath20.42 @xmath3 0.08 ) in si isotope composition , suggesting that secondary minerals in this sample do not measurably affect the isotopic composition of the bulk rock .
in contrast , the zeolite separate ( @xmath1 ^ 30^si = 0.00 @xmath3 0.08 ) is significantly heavier in si isotope composition than the mg-1001 whole - rock , so the zeolites are not the carrier of the light si isotope enrichment in the mangaia oib .
however , this is unexpected considering that fluid alteration generally leads to secondary phases enriched in light si isotopes .
the heavier si isotope composition of the zeolite can be explained as a precipitate that formed from si - bearing surface fluids that were unassociated with dissolution of the host rock ( i.e. , fluids that do not have @xmath1 ^ 30^si values matching those of basalts ) . assuming a typical surface fluid si isotope composition of 1.3 @xcite and considering that the si isotope fractionation factor between precipitated and dissolved si is around @xmath21 to @xmath21.5 for a variety of precipitates ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , the formation of secondary minerals with an isotopic composition around 0 is possible if the zeolite si source is dissolved si in such a fluid .
isotopically - heavy silica coatings have been observed in weathered hawaiian basalts and were interpreted as the result of weathering @xcite . since the zeolite fraction in mg-1001 has a relatively low modal abundance ( < 10% ) and low concentration of si ( < 4 wt.% ) , based on a simple mass balance the effect of the zeolite fraction on the bulk @xmath1 ^ 30^si value is less than 0.04 .
this would result in a positive @xmath1 ^ 30^si shift , suggesting that formation of secondary minerals due to weathering does not account for the light si isotope enrichment in oib lavas from mangaia . since equilibrium isotope fractionation decreases with increasing temperature ( e.g. , * ? ? ?
* ) , only a limited range of si isotope fractionation ( on the sub permil level ) is expected for high - temperature igneous processes . in general ,
evolved , silica - rich lithologies ( e.g. , rhyolites , dacites , granites ) have heavier si isotope compositions than less - evolved ( ultramafic and basaltic ) material , suggesting a small isotope fractionation during magmatic differentiation , with a @xmath00.2 offset in ^30^si/^28^si from 40 to 75 wt.% sio@xmath7 @xcite .
first - principles calculations predict equilibrium si isotope fractionation between co - existing mineral phases , with the more polymerized ( i.e. , more si - rich ) phase generally enriched in the heavier isotopes @xcite ; this suggests that si isotope fractionation may result from fractional crystallization or partial melting .
however , no significant si isotope fractionation has been observed as the result of basalt extraction by mantle partial melting @xcite .
^30^si as a function of sio@xmath7 content for the oib measured in this study .
no correlation between si isotope composition and silica content is observed , and widely variable @xmath1 ^ 30^si values are observed for a given sio@xmath7 value . a plot of @xmath1 ^ 30^si versus mgo or mg # shows similar results . ] the si isotope compositions of oib span a similar range ( i.e. , from the lightest oib at @xmath1 ^ 30^si = @xmath20.44 to the heaviest oib at @xmath1 ^ 30^si = @xmath20.22 ) as that defined by ultramafic to rhyolitic material in the literature ( e.g. , * ? ? ?
* ) , but since the sample set presented here consists of basaltic rocks , there is a much smaller range of sio@xmath7 contents ( table 1 ) .
furthermore , samples with similar sio@xmath7 contents ( e.g. , @xmath048 wt.%
sio@xmath7 ) can span a range of si isotope compositions up to 0.2 ( figure 4 ) . a plot of @xmath1 ^ 30^si versus mgo or mg # shows a similar lack of correlation with @xmath1 ^ 30^si ; this further argues against accumulation of isotopically light olivine and/or pyroxene as the carrier of low @xmath1 ^ 30^si values in oib .
the lack of correlation between @xmath1 ^ 30^si and sio@xmath7 content suggests that magmatic differentiation is not the primary cause of si isotope variability observed in oib .
recent theoretical calculations of equilibrium fractionation factors show that si isotopes may be fractionated between silicate minerals at the elevated pressure conditions in the lower mantle @xcite .
due to the large pressure gradients within the deep mantle , high - pressure silicates have a different structure than those at lower pressure ; low - pressure olivine polymorphs have four - fold coordination of si while high - pressure bridgmanite ( mg - perovskite ) has six - fold coordination of si . this difference in si coordination
changes the vibrational frequency associated with the si - o bonds , and isotopic substitution is sensitive to this effect . although isotopic fractionations are generally small at high - temperature , the work of @xcite predicts significant si isotope variations among silicate minerals with four - fold coordination of si ( e.g. , olivine , pyroxene , garnet ) and six - fold coordination of si ( bridgmanite ) . as a result , based on the calculated fractionation factors and estimates of the size and composition of mantle reservoirs ,
@xcite estimated the bridgmanite dominated lower mantle to be @xmath00.1 lighter on the ^30^si/^28^si ratio compared to the upper mantle .
if si isotope heterogeneity was created by bridgmanite crystallization at the base of the magma ocean during earth s silicate differentiation and subsequently preserved through mantle evolution , then the lower mantle may still be enriched in lighter si isotopes compared to the upper mantle .
although there is much debate about the internal structure of the mantle and the mechanisms responsible for the long - term preservation of primitive mantle reservoirs , layered convection models suggest that geochemical heterogeneities may be preserved for long periods ( e.g. , * ? ? ?
* ; * ? ? ?
the degree of variation in si isotope composition between oib measured in this study and previously measured morb derived from the shallow mantle ( e.g. , * ? ? ?
* ) is in agreement with the @xmath00.1 difference in si isotope composition between the upper and lower mantle predicted by @xcite , suggesting that preservation of si isotope heterogeneity within the mantle is a possible explanation for the @xmath1 ^ 30^si variability in oib .
the variable si isotope compositions of oib could potentially reflect selective sampling of preserved primordial components with light si isotope enrichment within the lower mantle . in the case of icelandic nrz oib
, their light si isotope enrichments may be consistent with sampling of a deep , primitive , undegassed mantle source based on the he and ne isotope systematics in these particular samples @xcite .
similarly , the so nicolau island of cape verde displays both relatively primitive ^3^he/^4^he ( 10 - 15 r / r@xmath6 ; * ? ? ?
* ) and a light si isotope enrichment compared to morb . arguments against light si isotope compositions reflecting a lower mantle signature include the observation that samoan ofu samples have primitive he @xcite and ne @xcite but morb - like si isotope composition ( @xmath1 ^ 30^si = @xmath20.31 @xmath3 0.03 ) .
conversely , canary island samples have ^3^he/^4^he values within the range of morb @xcite , but have variable si isotope compositions .
therefore , sampling of a primitive mantle reservoir with a light si isotope composition can not solely explain the si isotope variability in oib but remains a possibility . experimental work to constrain si isotope fractionation between silicates with different si coordination
are required to ground - truth the theoretical calculations of @xcite .
although only a limited number of samples in this study have previously been analyzed for o isotope compositions , it is noteworthy that most of the locations that exhibit light si isotope enrichments also have been identified as areas with low @xmath1 ^ 18^o signatures , including himu - type oib and iceland @xcite .
this raises the question of whether there is a common mechanism that generates the isotopically - light si and o isotope compositions of oib . like si isotopes , o isotopes display large variations in surface environments relative to mantle materials ; therefore , o isotopes can be a sensitive measure of low - temperature processes , particularly water - rock interactions .
altered oceanic crust can be enriched in the heavier isotopes of o through low - temperature alteration by surface fluids ( characteristic of the upper oceanic crust ) or depleted in the heavier isotopes through high - temperature hydrothermal alteration ( characteristic of the lower oceanic crust ) , so variations in @xmath1 ^ 18^o may indicate the presence of oceanic crustal material in oib magmas .
the depletions in the heavy isotopes of o found in some himu locations and iceland are consistent with either recycling or assimilation of hydrothermally altered lower oceanic crust @xcite .
it is difficult at this point to distinguish between recycling and assimilation of oceanic crust based solely on si isotopes .
however , due to the much lower concentration of dissolved si in fluids compared to the high o concentration , and the similar si isotope composition of si saturated hydrothermal fluids to their rock setting @xcite , the effect of hydrothermal alteration is likely to be much smaller for si isotopes than for o isotopes .
large amounts of hydrothermally altered material would likely be needed to create resolvable si isotope variations in oib .
based on the comparison of @xmath1 ^ 18^o and ^187^os/^188^os in oib , @xcite suggested that himu - type lavas compositions are inconsistent with the incorporation of recycled oceanic crust , and @xmath1 ^ 18^o depletions in oib reflect assimilation of lower oceanic crust .
however , @xcite argued that radiogenic ^187^os/^188^os coupled with high os concentrations in some canary island oib can not be explained by crustal assimilation , but rather indicates a contribution of recycled oceanic lithosphere .
this suggests that crustal assimilation is not the main cause of the light si enrichments in this oib suite , but processes of crustal recycling may play an important role in generating si isotope variability in the mantle .
subduction introduces a wide variety of materials into the mantle , including oceanic and continental crust , oceanic lithospheric mantle , and sediments , all of which may have undergone variable degrees of alteration on the seafloor and/or during subduction ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
subduction of geochemically diverse material that is incorporated into hotspot sources may give rise to observed heterogeneities in the products of these magmas , and it has been proposed that the type of material subducted generates the isotopic signatures of the different mantle components sampled by some oib @xcite . furthermore , subducted sediment is a significant source of incompatible elements to the mantle , and sediments may aid in creating chemical and isotopic heterogeneity within the mantle @xcite .
the presence in the mantle of unaltered continent - derived material ( in the form of lower continental crust or sediments derived from the upper continental crust that has not been altered in si isotope composition due to metamorphic dehydration and fluid - rock interaction during subduction ) is unlikely to significantly affect oib si isotope compositions .
estimates of the si isotope composition of both the lower and upper continental crust ( @xmath1 ^ 30^si = @xmath20.29 and @xmath1 ^ 30^si = @xmath20.25 , respectively ) are similar to morb and estimates for the bse @xcite .
although specific continental inputs may have si isotope compositions that differ from these values for the lower and upper continental reservoirs ( e.g. , shale , which exhibits @xmath1 ^ 30^si values from zero to @xmath20.82 ; * ? ? ?
* ) , we assume that on a large scale a single rock type would not dominate the subducted material and such sub - permil @xmath1 ^ 30^si variations would be averaged out . similarly , the presence of unaltered oceanic crust ( with morb - like si isotope composition ) would not create si isotope variations in oib .
however , isotopic variations in himu - like oib and the iceland nrz oib in this study have most often previously been explained by the incorporation of recycled oceanic crust into the plume source ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this suggests that if recycling of continental or oceanic crust is the source of the light si isotope enrichment observed in the himu and iceland nrz oib , the material must be geochemically modified either prior to or during subduction .
there are a number of processes acting on oceanic crust that could modify its si isotope composition , including high - temperature hydrothermal alteration at seafloor spreading centers , low - temperature seafloor alteration , and metasomatism and metamorphic dehydration during subduction @xcite .
the preferential loss of heavy si isotopes to the interacting fluid would result in isotopically light crustal material , which could lead to heterogeneities in the mantle as observed in oib .
although a high - precision investigation of the si isotope composition of altered and subducted lithologies ( e.g. , ophiolite sequences ) does not currently exist in the literature , there is some evidence for preservation of initial si isotope compositions during high - temperature metamorphism and metasomatism @xcite .
additionally , island arc basalts ( iab ) display si isotope compositions similar to morb @xcite , so processing during subduction may preserve primary si isotope compositions to the extent that the signature of interacting fluids is not communicated to the island arc .
altered oceanic crust displays enrichments in light si isotopes relative to morb @xcite .
it is therefore possible that oceanic crust has experienced si isotope fractionations due to seafloor and hydrothermal alteration , and this composition is preserved during subduction .
however , a systematic study of the behavior of si isotopes during metamorphic dehydration is necessary to ascertain the effects of subduction on mantle inputs .
in contrast to oceanic and continental crust , oceanic sediments have widely variable compositions and can be significantly enriched in light si isotopes compared to typical igneous rocks .
this suggests that the incorporation of oceanic sediments with isotopically light si in the oib plume sources is one viable mechanism for creating the light si isotope enrichments observed in oib .
previous measurements of si isotopes in silicified and precipitated materials such as cherts show a large range in si isotope composition , with @xmath1 ^ 30^si values down to @xmath24 or more @xcite , but sedimentary material with extremely fractionated si isotope compositions ( potentially including clays and cherts ) would be diluted by other subducting material .
additionally , much of the sedimentary material that enters subduction zones may be removed before entering the convecting mantle by forearc scraping or through arc volcanism @xcite ; in general , oib sources are believed to incorporate at most a few percent of sediments @xcite . for these reasons it is difficult to quantify the si isotope composition of subducting material that enters the mantle .
the potential contribution of subducted material on the si isotope signature of oib can be calculated from a simple mass balance using estimates for the si isotope composition and fraction of recycled material in the oib source .
here we consider a subducted package with a bulk @xmath1 ^ 30^si of @xmath21.0 , @xmath20.75 , and @xmath20.5 .
these values are consistent with plausible compositions of altered oceanic crust @xcite with or without a small portion of sediment ( we consider a few percent of sediments with @xmath1 ^ 30^si values down to @xmath24 corresponding to compositions of silicified and precipitated materials ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
results of the mass balance calculations are shown in figure 5 .
these calculations suggest that the relatively light si isotope enrichment observed in the himu and iceland nrz basalts can be explained by the incorporation 25% of recycled altered oceanic crust or 10 - 25% altered oceanic crust plus a few percent of sediments in oib magmas . using
a similar mass balance shows that 1 - 2% sediment added to peridotite mantle is unlikely to measurably shift the @xmath1 ^ 30^si of the mixture , which can explain the morb - like @xmath1 ^ 30^si values of em-1 and em-2 type oib despite their hosting some recycled marine sediment .
the addition of more than a few percent of sediments in the sources of oib , particularly himu - type , is inconsistent with pb and sr isotopic constraints and trace - element ratios unless sediments are significantly modified during subduction zone processing @xcite
. sediments constitute a thin veneer that overlies oceanic crust , so that the subduction package is dominated by basaltic oceanic crust and lithospheric mantle .
for this reason , it is more likely that the recycled material that can reproduce the @xmath1 ^ 30^si signatures of himu and iceland nrz oib is primarily altered ` basaltic ' oceanic crust with a bulk composition of @xmath1 ^ 30^si < @xmath20.5 that contributes melt to generate @xmath025% of the mass of the lavas .
furthermore , it should be noted that the mass fraction of recycled material in the mantle source is unlikely to be the same as the mass fraction present in the plume source melts because recycled lithosphere in the form of eclogite preferentially melts compared to ambient peridotitic mantle , potentially leading to pyroxenite - peridotite mixtures in oib mantle sources ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
^30^si as a function of the fraction of recycled component present in the melt derived from the mantle source .
the dashed line and shaded box represent the average si isotope composition ( @xmath3 2 se ) of the light - isotope enriched oib measured in this study ( iceland , mangaia , cape verde ) .
the bse reservoir is considered to have a si concentration [ si ] = 0.21 wt.% and @xmath1 ^ 30^si = @xmath20.29 @xcite .
the recycled component is modeled with a nearly identical bulk [ si ] = 0.23 wt.% but variable @xmath1 ^ 30^si values .
here @xmath20.50 , @xmath20.75 , and @xmath21.00 are used as illustrative @xmath1 ^ 30^si values for the bulk recycled component ; these values are consistent with altered oceanic crust @xcite or a mixture of altered oceanic crust with a few percent of sediments ( @xmath1 ^ 30^si values between @xmath22 and @xmath24 ; see text for details ) .
the si isotope compositions of the anomalous oib localities are consistent with the incorporation of 10 - 30% recycled material in the plume source melt . ] a key indication that the si isotope variability that potentially arises from recycling of subducted oceanic crust and lithospheric mantle lies in the difference in si isotope composition between lavas from the canary islands of la palma and el hierro .
these two islands have previously been found to have contrasting o isotope compositions , with relatively lower @xmath1 ^ 18^o values in la palma samples compared to el hierro samples @xcite .
furthermore , the coupled o and os isotope systematics of la palma and el hierro reflect different proportions of recycled oceanic crust and lithospheric mantle in the plume sources @xcite .
specifically , the relatively lower @xmath1 ^ 18^o values in la palma are generated by a contribution of recycled oceanic gabbroic crust @xcite .
this suggests that a single process may be responsible for generating the low @xmath1 ^ 18^o and low @xmath1 ^ 30^si values observed in oceanic basalts , although further work is necessary to confirm this relationship .
these observations indicate that the si isotope system may be an important tool for investigating the processes responsible for generating stable isotopic variability present in oib .
to a first order , oib exhibit homogeneous si isotope compositions generally in agreement with estimates for the @xmath1 ^ 30^si value of bse ( @xmath20.29 @xmath3 0.07 , 2 sd ; * ? ? ?
however , some systematic variations are present ; some himu - type and nrz basalts exhibit light si isotope enrichments relative to other oib , morb , and the estimate for the bse .
in contrast , em - type oib do not show any systematic si isotope variations .
one possible cause of the light si isotope enrichments in oib is primitive si isotope heterogeneity created by bridgmanite crystallization at the base of the magma ocean during the silicate differentiation of the earth as suggested by @xcite , although the potential to preserve such a reservoir over the history of the earth is debated .
an alternative explanation is that subduction of material that is incorporated into hotspot sources may give rise to observed heterogeneities .
the si isotope variability in the himu and iceland nrz can be explained by the presence of @xmath025% recycled altered oceanic crust and lithospheric mantle in the plume source .
a reservoir of isotopically light si in the earth s mantle would have implications for the calculated amount of si in the core .
silicon has long been proposed as a potential light element in the core to account for the density deficit relative to pure feni , and an estimate of the si content of the earth s core was one of the early applications of the si isotope system to high temperature geochemistry @xcite .
debate still persists about the degree of offset between the si isotope composition of the earth s mantle and that of chondrites ( assumed to represent the bulk earth composition ) , resulting in widely varying estimates of the si content of the earth s core from @xmath01 wt.% to more than 10 wt.% , although high amounts of si in the terrestrial core are likely irreconcilable with other geochemical and geophysical constraints ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , a reservoir of isotopically light si in the mantle may be indicated by these new measurements of oib .
this would lower the amount of si in the core required by mass balance , driving the calculated si content of the core based on si isotopes closer to that estimated by geophysical means .
we thank the three anonymous reviewers as well as the associate editor fangzhen teng for their thorough reviews and comments , which have greatly improved this paper .
ep thanks the chateaubriand stem fellowship program for funding .
fm thanks the european research council under the european community s h2020 framework program / erc grant agreement # 637503 ( pristine ) and the agence nationale de la recherche for a chaire dexcellence sorbonne paris cit ( idex13c445 ) and for the univearths labex program ( anr-10-labx-0023 and anr-11-idex-0005 - 02 ) .
ps thanks the support of the marie curie fp7-iof fellowship `` isovolc '' .
abraham k. , opfergelt s. , fripiat f. , cavagna a. , de jong j.t.m .
, foley s.f . , andr l. , cardinal d. , 2008 .
@xmath1 ^ 30^si and @xmath1 ^ 29^si determinations on usgs bhvo-1 and bhvo-2 reference materials with a new configuration on a nu plasma multi - collector icp - ms .
geostandards geoanalytical res .
32 , 193202 .
badro j. , fiquet g. , guyot f. , gregoryanz e. , occelli f. , antonangeli d. , dastuto m. , 2007 .
effect of light elements on the sound velocities in solid iron : implications for the composition of earth s core .
earth planet .
254 , 233238 .
brandenburg j.p . , hauri e.h . , van keken p.e .
, ballentine c.j . , 2008 . a multiple system study of the geochemical evolution of the mantle with force - balanced plates and thermochemical effects .
earth planet .
276 , 113 .
cabral r.a . ,
jackson m.g . , rose - koga e.f . ,
koga k.t . , whitehouse m. j. , antonelli m.a . , farquhar j. , day j.m.d .
, hauri e.h . , 2013 .
anomalous sulphur isotopes in plume lavas reveal deep mantle storage of archaean crust .
nature 496 , 490493 .
chemtob s.m . ,
rossman g.r .
, young e.d .
, ziegler k. , moynier f. , eiler j.m . , hurowitz j.a . , 2015 .
silicon isotope systematics of acidic weathering of fresh basalts , kilauea volcano , hawaii .
. cosmochim .
acta 169 , 6381 .
claude - ivanaj c. , joron j.l .
, allgre c.j . ,
^238^u^230^th^226^ra fractionation in historical lavas from the azores : long - lived source heterogeneity vs. metasomatism fingerprints .
176 , 295310 .
day j.m.d . , pearson d.g . ,
macpherson c.g . , lowry d. , carracedo j.c . , 2009 .
pyroxenite - rich mantle formed by recycled oceanic lithosphere : oxygen osmium isotope evidence from canary island lavas .
geology 37 , 555558 .
day j.m.d . ,
pearson d.g . ,
macpherson c.g . , lowry d. , carracedo j.c . , 2010 .
evidence for distinct proportions of subducted oceanic crust and lithosphere in himu - type mantle beneath el hierro and la palma , canary islands .
geochim .
cosmochim .
acta 74 , 65656589 .
fitoussi c. , bourdon b. , kleine t. , oberli f. , reynolds b.c .
si isotope systematics of meteorites and terrestrial peridotites : implications for mg / si fractionation in the solar nebula and for si in the earth s core .
earth planet .
287 , 7785 .
garapic g. , jackson m.g .
, hauri e.h . ,
hart s.r . ,
farley k.a .
, blusztajn j.s . , woodhead j.d . , 2015 . a radiogenic isotopic ( he sr
os ) study of lavas from the pitcairn hotspot : implications for the origin of em-1 ( enriched mantle 1 ) .
lithos 228229 , 111 .
geist d.j . ,
fornari d.j . ,
kurz m.d . ,
harpp k.s .
, adam soule s. , perfit m.r .
, koleszar a.m. , 2006 .
submarine fernandina : magmatism at the leading edge of the galapagos hot spot .
geochem .
geophys .
geosyst . 7 , q12007 . hanyu t. , kawabata h. , tatsumi y. , kimura j .- i .
, hyodo h. , sato k. , miyazaki t. , chang q. , hirahara y. , takahashi t. , senda r. , nakai s. , 2014 .
isotope evolution in the himu reservoir beneath st .
helena : implications for the mantle recycling of u and th .
geochim . cosmochim .
acta 143 , 232252 .
hendry k.r . , georg r.b .
, rickaby r.e.m . ,
robinson l.f .
, halliday a.n . , 2010 .
deep ocean nutrients during the last glacial maximum deduced from sponge silicon isotopic compositions .
earth planet .
292 , 290300 .
herzberg c. , cabral r.a .
, jackson m.g .
, vidito c. , day j.m.d .
, hauri e.h . , 2014 .
phantom archean crust in mangaia hotspot lavas and the meaning of heterogeneous mantle .
earth planet .
396 , 97106 .
huang s. , farkas j. , jacobsen s.b . , 2011 .
stable calcium isotopic compositions of hawaiian shield lavas : evidence for recycling of ancient marine carbonates into the mantle .
geochim .
cosmochim .
acta 75 , 49874997 .
jackson m.g . , kurz m.d . ,
hart s.r . , 2009 .
helium and neon isotopes in phenocrysts from samoan lavas : evidence for heterogeneity in the terrestrial high ^3^he/^4^he mantle .
earth planet .
287 , 519528 .
kawabata h. , hanyu t. , chang q. , kimura j .- i . ,
nichols a.r.l . , tatsumi y. , 2011 .
the petrology and geochemistry of st .
helena alkali basalts : evaluation of the oceanic crust - recycling model for himu oib . j. pet . 52 , 791838 .
macpherson c.g .
, hilton d.r . ,
day j.m.d .
, lowry d. , gronvold k. , 2005 .
high-^3^he/^4^he , depleted mantle and low-@xmath1 ^ 18^o , recycled oceanic lithosphere in the source of central iceland magmatism .
earth planet .
233 , 411427 .
mheut m. , lazzeri m. , balan e. , mauri f. , 2007 .
equilibrium isotopic fractionation in the kaolinite , quartz , water system : prediction from first - principles density - functional theory .
geochim . cosmochim .
acta 71 , 31703181 .
millet m .- a .
, doucelance r. , schiano p. , david k. , bosq c. , 2008 .
mantle plume heterogeneity versus shallow - level interactions : a case study , the so nicolau island , cape verde archipelago .
j. volcanol .
geotherm .
176 , 265276 .
opfergelt s. , georg r.b .
, delvaux b. , cabidoche y .- m .
, burton k. w. , halliday a.n . , 2012 .
silicon isotopes and the tracing of desilication in volcanic soil weathering sequences , guadeloupe .
geol . 326327 , 113122 .
rickers f. , fichtner a. , trampert j. , 2013 .
the iceland
jan mayen plume system and its impact on mantle dynamics in the north atlantic region : evidence from full - waveform inversion .
earth planet .
367 , 3951 .
savage p.s . ,
georg r.b .
, williams h.m .
, halliday , a.n .
silicon isotopes in granulite xenoliths : insights into isotopic fractionation during igneous processes and the composition of the deep continental crust .
earth planet .
365 , 221231 .
savage p.s .
, wimpenny j. , harvey j. , yin q. , moynier f. , 2013c .
investigating the effects of abyssal peridotite alteration on si , mg and zn isotopes .
abstract v54a-06 presented at 2013 fall meeting , agu , san francisco , ca , 913 december .
sobolev a.v . ,
hofmann a.w .
, kuzmin d.v . ,
yaxley g.m . ,
arndt n.t . ,
chung s.l . ,
danyushevsky l.v .
, elliott t. , frey f.a . , garcia m.o . ,
gurenko a.a . ,
kamenetsky v.s . , kerr a.c . ,
krivolutskaya n.a . ,
matvienkov v.v . ,
nikogosian i.k .
, rocholl a. , sigurdsson i.a . , sushchevskaya n.m . and teklay m. , 2007 .
the amount of recycled crust in sources of mantle - derived melts .
science 316 , 412417 .
steinhoefel g. , von blanckenburg f. , horn i. , konhauser k.o .
, beukes n.j . and gutzmer j. , 2010 .
deciphering formation processes of banded iron formations from the transvaal and the hamersley successions by combined si and fe isotope analysis using uv femtosecond laser ablation .
geochim .
cosmochim .
acta 74 26772696 .
thirwall m.f . ,
gee m.a.m .
, lowry d. , mattey d.p .
, murton b.j . ,
taylor r.n . , 2006 .
low @xmath1 ^ 18^o in the icelandic mantle and its origins : evidence from reykjanes ridge and icelandic lavas .
geochim .
cosmochim .
acta 70 , 9931019 .
van den boorn s.h . , van bergen m.j . ,
vroon p.z .
, de vries s.t . , nijman w. , 2010 .
silicon isotope and trace element constraints on the origin of @xmath03.5 ga cherts : implications for early archaean marine environments .
geochim . cosmochim .
acta 74 , 10771103 .
woodhead j.d . ,
mcculloch m.t . , 1989 .
ancient seafloor signals in pitcairn island lavas and evidence for large amplitude , small length - scale mantle heterogeneities .
earth planet .
94 , 257273 .
workman r.k .
, hart s.r . , jackson m. , regelous m. , farley k.a . , blusztajn j. , kurz m. , staudigel h. , 2004 .
recycled metasomatized lithosphere as the origin of the enriched mantle ii ( em2 ) end - member : evidence from the samoan volcanic chain .
geochem .
geophys .
geosys . 5 , q04008 . zambardi t. , poitrasson f. , corgne a. , mheut m. , quitt g. , anand m. , 2013 .
silicon isotope variations in the inner solar system : implications for planetary formation , differentiation and composition .
geochim .
cosmochim .
acta 121 , 6783 . | the study of silicon ( si ) isotopes in ocean island basalts ( oib ) has the potential to discern between different models for the origins of geochemical heterogeneities in the mantle . relatively large ( @xmath0several per mil per atomic mass unit )
si isotope fractionation occurs in low - temperature environments during biochemical and geochemical precipitation of dissolved si , where the precipitate is preferentially enriched in the lighter isotopes relative to the dissolved si .
in contrast , only a limited range ( @xmath0tenths of a per mil ) of si isotope fractionation has been observed from high - temperature igneous processes .
therefore , si isotopes may be useful as tracers for the presence of crustal material within oib mantle source regions that experienced relatively low - temperature surface processes in a manner similar to other stable isotope systems , such as oxygen .
characterizing the isotopic composition of the mantle is also of central importance to the use of the si isotope system as a basis for comparisons with other planetary bodies ( e.g. , moon , mars , asteroids ) . here
we present the first comprehensive suite of high - precision si isotope data obtained by mc - icp - ms for a diverse suite of oib .
samples originate from ocean islands in the pacific , atlantic , and indian ocean basins and include representative end - members for the em-1 , em-2 , and himu mantle components . on average ,
@xmath1 ^ 30^si values for oib ( @xmath20.32 @xmath3 0.09 , 2 sd ) are in general agreement with previous estimates for the @xmath1 ^ 30^si value of bulk silicate earth ( @xmath20.29 @xmath3 0.07 , 2 sd ; * ? ? ?
* ) .
nonetheless , some small systematic variations are present ; specifically , most himu - type ( mangaia ; cape verde ; la palma , canary islands ) and iceland oib are enriched in the lighter isotopes of si ( @xmath1 ^ 30^si values lower than morb ) , consistent with recycled altered oceanic crust and lithospheric mantle in their mantle sources . |
the milky way appears to be a typical spiral galaxy with a disk and bulge .
while our unique inside view of the galaxy helps to understand the galactic structure in general , it also makes it more difficult to identify structures such as bars . the case for existence of a bar at the galactic centre first proposed by @xcite is easier to make knowing that bars are common in external galaxies .
there is now conclusive evidence that the galactic bulge is of a barred type .
the longitude asymmetry of the cobe photometric maps ( @xcite ) , high optical depths to gravitational microlensing ( @xcite ) , asymmetric star counts ( @xcite ) , non - circular gas kinematics ( @xcite ) , and triaxiality of the stellar velocity field ( @xcite ) have all been interpreted as signatures of the galactic bar .
unfortunately , the size and precise orientation of the bar are still being debated .
recently @xcite found that the infra - red star counts collected by the spitzer space telescope are best explained assuming a bar with a half - length @xmath10 kpc placed at a @xmath11 angle to the sun galactic center line .
most previous studies prefer a bar at @xmath12 extending out to @xmath93.5 kpc ( e.g. @xcite ) .
such apparently conflicting evidence may be an indication that the inner galaxy hides even more complicated structures . a secondary bar ( @xcite ) and a ring ( @xcite )
have been suggested , since these features are also evident in many other spiral galaxies ( e.g. @xcite ) .
@xcite discussed the progress of the dynamical modeling techniques in the context of major observational advances expected from a future space mission gaia .
the three approaches to constructing a self - consistent dynamical galaxy model are the schwarzschild method ( @xcite ) , the torus modelling method ( see @xcite for details ) and @xmath13-body simulations with particle weights determined by the `` made - to - measure '' algorithm ( @xcite ) .
the first galactic bar model employing the latter method was built by @xcite .
neither of these techniques can fully address the structure of the inner galaxy without constraints on stellar kinematics .
the refinement of the models is limited largely by the scarcity of good proper motion and radial velocity measurements .
@xcite , for example , compared kinematic predictions of their model with the data for just two lines of sight .
a handful of samples published since the pioneering photographic work of @xcite is not enough to remove the non - uniqueness of the model parameters . in a study based on two lines of sight @xcite
have demonstrated that high quality relative proper motions can be obtained with a relatively modest investment of time using the hubble space telescope ( hst ) . at the resolution of the wide field planetary camera 2 ( wfpc2 ) instrument the required time baseline is only a few years .
the hst archive contains a number of images suitable as the first epoch data , so the tedious part of accumulating the baseline can be avoided entirely .
most of these fields are centred around microlensing events discovered by the macho collaboration ( e.g. @xcite ) . using a similar concept to that of @xcite
, we carried out a mini - survey of proper motions in 35 of the available macho fields to study the kinematics of microlensed sources and of the general stellar populations . here
we present the measurement techniques and results for the general stellar population in these 35 lines of sight .
[ cols="<,^,^,^,<,<,^,^,^ , < " , ] [ tab : krcomp ]
the study of @xcite focused on cleaning the galactic bulge population and removing the contamination by the bluer disk stars . above the bulge turn - off point ,
the stellar colors alone are sufficient to separate the blue disk main sequence from the red giants , subgiants and clump giants .
the size of our fields is generally too small to provide useful statistics of bright stars above turn - off point , and good color information is only available for about 1/3 of the lines of sight .
however , three of the fields in table [ tab : data ] with useful colors ( 97-blg-18 , 104-c and 104-d ) are close to each other and were combined in order to look for a kinematic distinction between the galactic disk and bulge populations .
figs [ fig : cmd_pm_dist ] and [ fig : pm_dist ] show that in the frame of reference of a mean star ( of any color ) , the longitude proper motions of the blue disk stars are biased toward positive values , while the red bulge stars tend to have more negative @xmath14 .
the blue and red samples were selected , correspondingly , using conditions @xmath15 and @xmath16 .
this effect was previously observed by @xcite and by @xcite in their two fields with multi - epoch wfpc2 data and there is little doubt that it is due to the blue disk stars `` rotating in front '' of the red bulge stars .
@xcite also devised an approximate distance measure : @xmath17 chosen to remove the slope of the main sequence in the color - magnitude diagram . in fig .
[ fig : kr2 ] we present the average proper motions and their dispersions for the baade window in bins of @xmath18 .
as expected , with an increasing depth along the line of sight , the kinematic signature gradually changes from that characteristic of the disk stars , to the one typical for the bulge . in the @xcite data this trend continues to very faint stars that are likely on the far side of the bulge , and
if so , it constitutes a `` rotation curve '' of the bulge .
the colors for our fields are generally of lower s / n ratio or nonexistent , and do not allow to see this much detail .
a detailed modeling of the measurements in table [ tab : pm_results ] is beyond the scope of this paper .
here we only comment on possible directions and new possibilities .
@xcite interpreted the bulge anisotropy in terms of the rotation support of the galactic bulge and related the ratio @xmath7 directly to the level of flattening of the light density distribution .
they also concluded that the value @xmath191.15 observed in baade window @xmath20 , with which our measurements are consistent , can be explained by rigid rotation .
the presence of any disk stars , however , will also contribute rotational broadening to @xmath5 .
since in the vicinity of our fields the disk fraction increases closer to the plane , it follows that the measured gradient from equation ( [ eq : sigl_b ] ) could be due to the disk contamination .
the changes of skewness in the @xmath14 distribution tend to support this ( see fig .
[ fig : pm_dist ] ) .
another possibility is that the rotation rate of the bulge actually increases at lower @xmath21 , as found by @xcite from the radial velocities of 124 sio masers in the galactic bulge .
it has been observed that for giants in baade window the metal - poor stars display more spread in the vertical motion and less anisotropy when compared to metal - rich samples ( @xcite ) .
both of these metallicity dependencies are quite steep , so it is likely that the gradient from equation ( [ eq : sigb_l ] ) is related to a changing mix of populations with more metal - poor stars closer to the galactic bulge minor axis .
we are not aware of any previous detections of the cross terms in the galactic bulge velocity field except the report by @xcite of a significant vertex deviation between the radial and longitudinal motions from @xmath22 .
that result is based on a photographic sample of @xmath9200 k and m giants from @xcite .
we note that the latter sample actually shows a hint of a slightly negative covariance between @xmath14 and @xmath23 ( c.f . fig . 1 of @xcite ) .
the superb resolution of hst enabled very significant detections of the @xmath24 cross term in many fields .
the non - diagonal elements of the velocity tensor are crucial to determining the dominant orbit families , the importance of streaming motions and the need for the intrinsic anisotropy versus solid body rotation in the galactic bulge ( @xcite ) .
@xcite published detailed calculations of @xmath24 for several lines of sight at positive longitudes including baade window @xmath25 , and two other : @xmath26 and @xmath27 .
taken at face value these predictions are roughly of the same magnitude as the results from [ sec : results ] , but have the opposite sign . for a proper comparison with dynamical models like the ones in @xcite and
@xcite we need to wait until the calculations are folded with the appropriate selection functions , since our measurements are based on substantially deeper data than most of the previous samples .
the main results of our proper motion mini - survey are : ( 1 ) high quality proper motion measurements for hundreds of stars in 35 lines of sight across the galactic bar , ( 2 ) establishing the presence of spatial gradients in dispersions @xmath28 and the amount of anisotropy @xmath7 , and ( 3 ) the first reliable detection of the covariance term @xmath24 of the transverse velocity tensor .
we cross - validated our measurements with the ground based ogle - ii data of @xcite and a benchmark study of @xcite . the observed slow rise of @xmath5 toward the galactic plane is likely due to the increasing disk contamination and/or a possible gradient in the bulge rotation speed .
the increase in @xmath6 toward the minor axis of the bulge is accompanied by the decreasing ratio @xmath7 and possibly results from increasing fraction of metal - poor stars .
another possibility is that @xmath6 increases due to the larger surface density of stars at low @xmath29 ( closer to the galactic centre ) .
we clearly detect the covariance term @xmath30 that implies a significant tilt of the galactic bulge velocity ellipsoid with respect to the galactic plane . using the same procedures as in @xcite , we find the tilt is roughly equal to @xmath31 .
the data presented in this paper provide qualitatively new constraints on dynamical models of the inner galaxy and dramatically improved number statistics .
furthermore , it may be possible in the near future to augment our proper motion samples with the distance and metallicity estimates . as shown by @xcite
, deep color - magnitude diagrams can supply sufficiently accurate distance information to effectively isolate the bulge population . in order to maximize the discriminating power of model comparisons
the focus should be on extending the coverage to negative longitudes and locations further from the galactic centre .
we thank profs .
bohdan paczyski , ian browne and drs .
vasily belokurov , wyn evans and nicholas rattenbury for helpful comments .
support to p.w . for proposal snap-10198 was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
this work was partly supported by the european community s sixth framework marie curie research training network programme , contract no .
mrtn - ct-2004 - 505183 `` angles '' , in particular s.k . through a studentship , l.w . through a pdra , and s.m . through travel support .
alard , c. , 2001 , a&a , 379 , l44 alcock , c. , et al .
, 2000 , apj , 541 , 734 babusiaux , c. , gilmore , g. , 2005 , mnras , 358 , 1309 benjamin , r. a. , et al . , 2005 , apj , 630 , l149 binney , j. j. , 2005 , preprint ( astro - ph/0411229 ) binney , j. j. , merrifield , m. r. , 1998 , galactic astronomy ( princeton university press ) bissantz , n. , debattista , v. p. , gerhard , o. , 2004 , apj , 601 , l155 blitz , l. , spergel , d. n. , 1991 , apj , 379 , 631 de vaucouleurs , g. , 1964 , in kerr , f. , rodgers , a. , eds , proc .
20 , the galaxy and the magellanic clouds .
australian academy of sciences , canberra , p. 195 dwek , e. , et al .
, 1995 , apj , 445 , 716 erwin , p. , sparke , l. s. , 1999 , apj , 521 , l37 fruchter , a. s. , hook , r. n. , 2002 , pasp , 114 , 144 gerhard , o. , 2001 , in asp conf . ser .
230 , galaxy disks and disk galaxies , eds . j. g. funes , & e. m. corsini
( san francisco : asp ) , 21 girardi , l. et al . , 2002 ,
a&a , 391 , 195 groth , e. j. , 1986 , aj , 91 , 1244 hfner , r. , evans , n. w. , dehnen , w. , binney , j. , 2000 , mnras , 314 , 433 izumiura , h. , et al . , 1995 , apjs , 98 , 271 koekemoer , a. m. , fruchter , a. s. , hook , r. n. , hack , w. , the 2002 hst calibration workshop , 2002 , 339 kuijken , k. , rich , r. m. , 2002 , aj , 124 , 2054 popowski , p. , et al . , 2005 ,
apj , 631 , 879 sevenster , m. n. , kalnajs , a. j. , 2001 , aj , 122 , 885 spaenhauer , a. , jones , b. f. , whitford , a. e. , 1992 , aj , 103 , 297 sumi , t. , et al . , 2004 , mnras , 348 , 1439 sumi , t. , et al . , 2005 , mnras , 356 , 331 syer , d. , tremaine , s. , 1996 , mnras , 282 , 223 stanek , k. z. , mateo , m. , udalski , a. , szymaski , m. , kaluzny , j. , & kubiak , m. , 1994 , apj , 429 , l71 udalski , a. , kubiak , m. , szymaski , m. , 1997 , acta astronomica , 47 , 319 woniak , p. , 2000 , acta astronomica , 50 , 421 zhao , h.s . , spergel , d. n. , rich , r. m. , 1994 , aj , 108 , 2154 zhao , h.s . , spergel , d. n. , rich , r. m. , 1995 , apj , 440 , l13 zhao , h.s . ,
rich , r. m. , biello , j. , 1995 , apj , 470 , 506 | we present a proper motion mini - survey of 35 fields in the vicinity of baade window , @xmath0 , sampling roughly a @xmath1 deg region of the galactic bar .
our second epoch observations collected with the acs / hrc instrument on board the hubble space telescope were combined with the archival wfpc2/pc images .
the resulting time baselines are in the range of 48 years
. precise proper motions of 15,863 stars were determined in the reference frame defined by the mean motion of stars with magnitudes between @xmath2 along the line of sight .
we clearly detect small gradients in proper motion dispersions @xmath3 , and in the amount of anisotropy ( @xmath4 ) .
both the longitude dispersion @xmath5 and its ratio to the vertical motion @xmath6 increase toward the galactic plane .
the decline of the anisotropy ratio @xmath7 toward the minor axis of the bulge is mostly due to increasing @xmath6 .
we also find , for the first time , a significant negative covariance term in the transverse velocity field @xmath8 .
our results extend by a factor of @xmath915 the number of the galactic bar fields with good proper motion dispersions .
galactic dynamics - stars : proper motion , dispersion + gravitational lensing - galaxy : bar , bulge , disc |
in this work we present a derivation of the fixed center approximation to the faddeev equations for the @xmath0 system for the case when the @xmath4 system has total spin @xmath1 .
the case for @xmath5 was done in @xcite for @xmath6 at threshold and extended to values below threshold to find a quasibound state in @xcite . an earlier version for @xmath7
was done in @xcite but ignored charge exchange processes in the multiple scattering of the kaons .
the improved results of @xcite for @xmath8 provided a binding of about 12 mev smaller than ignoring the charge exchange processes and was worth being taken into consideration .
the resulting binding was 9 mev with a width of 30 mev . yet , a formula like the one of @xcite for the case of @xmath1 was never derived and one is left with some worry about what could be the effects of charge exchange processes in this latter case .
this is the purpose of the present paper , where a simple derivation is presented and the new results are shown .
the study of the bound @xmath0 system has been the center of much attention recently @xcite .
earlier calculations on this system were already done in the 60 s @xcite and more recently in @xcite , yet , the works of @xcite improved considerably on these earlier works .
two different methods have been used in those papers .
those of refs .
@xcite use faddeev equations , in the formulation of alt - grassberger - sandhas ( ags ) @xcite , using a separable interaction for the potentials with energy independent strength , and form factors depending on the three momenta .
they include as coupled channels @xmath9 and @xmath10 , the basic ones in the study of the @xmath11 interaction in chiral unitary theories at low energies @xcite .
on the other hand the works of @xcite use a variational method to obtain the binding energy and width , by means of an effective potential @xcite that leads to the strongly energy dependent @xmath12 amplitudes of the chiral unitary approach .
this includes the @xmath13 channel integrated into the effective @xmath12 potential , but do not include the @xmath14 channel explicitly into the three body system . in table iv of @xcite one finds that there are 11 mev difference considering just one channel @xmath12 or the two channels @xmath12 and @xmath13 , both in the evaluation of the @xmath12 amplitude and in the @xmath15 system , finding more binding in the case of two channels .
one should note that this is not to be compared with the approach of @xcite where the @xmath12 amplitude is always calculated with @xmath12 , @xmath13 and other coupled channels .
one may argue that the consideration of @xmath13 in the two body system would account for most of the @xmath13 influence in the three body problem , a conjecture made by schick and gibson @xcite which was substantiated numerically in the work of ref .
but this was only done at threshold of @xmath11 , so testing this at lower energies is worthwhile and we shall face this problem here too .
the two faddeev approaches @xcite lead to binding energies higher than the variational approach , 50 - 70 mev versus around 20 mev binding respectively .
the use of an energy independent kernel in the ags equations is partly responsible for the extra binding of these approaches with respect to the chiral calculations .
the chiral potential is energy dependent , proportional to the sum of the two external meson energies in @xmath16 , and , as a consequence , one obtains a smaller @xmath16 amplitude at @xmath17 energies below threshold , resulting in a smaller binding for the @xmath18 system @xcite .
this is also corroborated in the approach of @xcite when the energy dependence of the weinberg - tomozawa chiral potential is taken into account in @xcite .
further discussions along this issue can be found in section 5 of @xcite . yet , in most cases the widths are systematically larger than the binding energy , of the order of 70 - 100 mev , which makes the observation of these states problematic . a discussion on experimental work claiming to have found a @xmath19 bound state is made in section 5 of @xcite , together with the theoretical work that shows that the experimental peaks observed correspond to conventional mechanisms that have nothing to do with a new bound state .
the approach of @xcite follows the chiral unitary approach for the @xmath20 amplitudes , which provide the most important source of the binding of the three body system .
however , rather than looking for the binding , searching for poles in the complex plane or looking for the energy that minimizes the expectation value of the hamiltonian as the other approaches do , one looks for peaks in the scattering matrices as a function of the energy of the three body system .
these amplitudes could in principle be used as input for final state interaction when evaluating cross sections in reactions where eventually this @xmath21 state is formed .
we rely upon the results of @xcite which prove that only on shell two body amplitudes are needed to study the three body system . by this
we mean the part of the analytical amplitudes obtained setting @xmath23 for the external particles , even if the particles are below threshold . like in the other works
, we also assume that the two body interactions proceed in l=0 .
according to all the works , the main component of the wave function corresponds to having a @xmath24 in i=0 ( coupling strongly to the @xmath25 and hence the total isospin will be i=1/2 .
the spin of the system for l=0 is provided by the two nucleons and all the works find that the most bound case corresponds to @xmath1 . since the @xmath5 case was studied in @xcite , here we concentrate in the @xmath1 case . since in our case @xmath26 corresponds to the total spin of the @xmath18 system we shall refer to @xmath27 in what follows . in the fca
the @xmath18 system is addressed by studying the interaction of a @xmath2 with a @xmath4 cluster .
the scattering matrix for this system is evaluated as a function of the total energy of the @xmath18 system and one looks for peaks in @xmath28 . in a second step one
allows explicitly the intermediate @xmath29 state in the three body system .
a discussion of different approaches along the fca line is done in @xcite , where it is shown that some works do not take into account explicitly the charge exchange @xmath30 reactions and antisymmetry of the nucleons , but it is explicitly done in ref .
@xcite . in @xcite charge exchange in the @xmath2 multiple scattering
was also ignored , but in @xcite it was taken into account for the @xmath5 case , and a quasibound state was found with about 9 mev binding and a width of about 30 mev .
technically we follow closely the formalism of @xcite , where the fca has been considered , using chiral amplitudes , in order to study theoretically the possibility of forming multi-@xmath31 states with large spin .
the @xmath4 interaction is of long range and very strong .
it binds the deuteron in spin @xmath5 and @xmath32 , with @xmath33 to a very good approximation .
it nearly binds the @xmath34 system also .
the binding of the @xmath34 system is so close that a little help from an extra interaction , the strongly attractive @xmath35 interaction , is enough to also bind this system and we shall assume that this is the case here for @xmath34 or in general for two nucleons in @xmath1 , @xmath36 , @xmath33 .
the strategy followed in @xcite is to assume as a starting point that the @xmath4 cluster has a wave function like the one of the deuteron ( we omit the d - wave ) . later on one
releases this assumption and assumes that the @xmath4 system is further compressed in the @xmath37 system . for this
we rely upon the information on the @xmath4 radius in the @xmath37 molecule obtained in @xcite , where the @xmath4 interaction is taken into account including short range repulsion that precludes unreasonable compression .
the r.m.s radius found is of the order of 2.2 fm , slightly above one half the value of the deuteron r.m.s radius of 3.98 fm ( @xmath4 distance ) .
in order to evaluate the amplitude for the @xmath18 with @xmath1 , we follow the following strategy .
\1 ) we evaluate first the @xmath38 amplitude considering charge exchange processes in the rescattering of the kaons .
the amplitude will contains total isospin @xmath39 and @xmath40 .
since we only want the @xmath39 , we evaluate the scattering amplitude for @xmath40 in addition , taking the @xmath41 amplitude , and from a linear combination of the two we obtain the @xmath39 amplitude .
we have found that this strategy leads to a far simpler derivation than other methods that we have also tried .
\2 ) for the @xmath38 we define three partition functions , \a ) @xmath42 , which contains all diagrams that begin with a @xmath43 collision with the first proton of the @xmath34 system and finish with @xmath44 .
\b ) @xmath45 , which contains all the diagrams that begin with a @xmath46 collision on a @xmath47 system and finish with @xmath44 .
\c ) @xmath48 , which contains all the diagrams that begin with a @xmath49 collision on a neutron and finish with @xmath50 .
these amplitudes fulfill a set of coupled equations that are easy to see diagrammatically by looking at the diagrams of fig . [
fig : tpp22 ] .
.,title="fig:",scaledwidth=100.0% ] + .,title="fig:",scaledwidth=60.0% ] + .,title="fig:",scaledwidth=100.0% ] we have : @xmath51 where @xmath52 , @xmath53 , @xmath54 , @xmath55 and @xmath56 is the function @xcite @xmath57 where @xmath58 is the form factor of the @xmath4 system .
we can see that eq .
( [ eq : gzero ] ) contains the folding of the @xmath2 intermediate propagator with the form factor of the @xmath4 system .
the value of the @xmath59 in the total rest frame is given by @xmath60 the argument of the @xmath61 , @xmath62 amplitudes is obtained following the steps of @xcite and is given by @xmath63 by taking into account that @xmath64 in the isospin basis , we can write all the former elementary amplitudes in terms of @xmath65 ( @xmath66 ) for the @xmath12 system , and we find @xmath67 solving eq .
( [ eq : tptexp ] ) we find @xmath68 which in isospin basis can be simplified to @xmath69 the total @xmath38 amplitude would be @xmath70 accounting for the first interaction of the @xmath43 with either of the protons .
\3 ) now we take into account that in the basis of @xmath71 @xmath72 and thus @xmath73 the @xmath74 amplitude is particularly easy to obtain . in this case
we take the @xmath41 transition and diagrammatically we have the mechanism of fig .
[ tpp3b2 ] for the only partition function @xmath75 .
hence @xmath76 and the total @xmath77 amplitude will be @xmath78 , accounting for the @xmath49 interacting first also with the second nucleon .
we have @xmath79 we can now use eq .
( [ eq : t1b2 ] ) and find for the total amplitude ( including the factor two for first interaction with either proton ) @xmath80 which can be simplified dividing the numerator by @xmath81 with the final result @xmath82 this equation improves over the one obtained in @xcite although they coincide for first scattering term .
the @xmath13 channel and other coupled channels are explicitly taken into account in the consideration of the @xmath12 amplitude which we have used in the fca approach .
this means that we account for the @xmath84 transition , and an intermediate @xmath83 channel , but this @xmath85 state is again reconverted to @xmath12 leaving the other @xmath86 as a spectator .
this is accounted for in the multiple scattering in the @xmath87 scattering matrix on one nucleon .
however , we do not consider the possibility that one has @xmath84 and the @xmath88 rescatters with the second nucleon . if we want to have a final @xmath21 system again , the @xmath88 has to scatter later with the @xmath89 to produce @xmath12 .
one may consider multiple scatterings of the @xmath88 with the nucleons , but given the smallness of the @xmath90 amplitude compared to the @xmath12 , any diagram having more than one rescattering of the pion with the nucleon will be negligible .
this means that to account for explicit @xmath91 state in the three body system it is sufficient to consider the diagram of fig .
[ fig:10 ] ( for @xmath12 scattering on nucleon 1 ) . channel .
the equivalent diagrams where the @xmath92 interacts first with the second nucleon should be added.,scaledwidth=25.0% ] the explicit expression for this contribution can be seen @xcite ( section iv ) .
in particular it involves the isospin combination in the @xmath93 @xmath94 we note , in passing , that the combination found for @xmath8 was @xmath95 and the @xmath96 amplitudes were also different .
this justifies the different effects found for these two cases , although the changes are small in both of them . non - perturbatively we can add @xmath97 of @xcite to eq .
( [ eq : isospins0 ] ) . in as much as the term
@xmath98 is the largest component of eq .
( [ eq : tknncharge ] ) , if we add @xmath97 to @xmath98 in all terms in eq .
( [ eq : tknncharge ] ) , we guarantee the correct results for the lowest order contribution and account approximately for multiple scattering including the @xmath99 contribution .
we have seen that the perturbative and non - perturbative results are very similar , but there is a difference of about 2 mev in the position of the peak ( the non - perturbative case more bound ) .
the solid line indicates the @xmath18 system with partial omission of charge exchange @xcite , the dashed line indicates the same system taking into account charge exchange diagrams and the dot - dashed line indicates the result of the including the @xmath91 channel and charge exchange diagrams.,scaledwidth=100.0% ] in figs .
[ fig : s0 ] and [ fig : s0red ] we show our results . in fig .
[ fig : s0 ] we use the form factor of the deuteron for the @xmath4 system and we show the results for @xmath28 in three options : a ) partial omission of charge exchange , using the expression obtained in @xcite @xmath100 by partial omission we mean that some charge exchange is accounted for when making the transition in the single scattering from @xmath101 to @xmath102 , the two components of the @xmath103 wave function .
however , charge exchange in double ( in multiple ) scattering in the diagonal transitions @xmath104 would be neglected .
this is unlike the @xmath105 system where the @xmath106 transition in single scattering is not allowed ; b ) charge exchange explicitly included , eq .
( [ eq : tknncharge ] ) ; c ) explicit @xmath91 channel added to the amplitude of eq .
( [ eq : tknncharge ] ) , as we have described in the former section .
size from @xcite .
the solid line indicates the @xmath18 system with partial omission of charge exchange , the dashed line indicates the same system taking into account charge exchange diagram and the dot - dashed line indicates the result of the including the @xmath91 channel and charge exchange diagrams.,scaledwidth=100.0% ] we can see from the figure that the consideration of charge exchange processes reduces the binding in 4 mev with respect to the results in @xcite .
this is smaller than the 12 mev shift found in @xcite for the @xmath5 case .
also , relatively , the change is more drastic in the @xmath5 case since there was a reduction of 12 mev from 21 mev , while here there is a reduction of 4 mev over a total of 32 mev binding .
when we consider the explicit intermediate @xmath91 state we find an extra reduction of 1 mev in the binding .
the results are similar , but the differences more pronounced , when we redo the calculations using the reduced @xmath4 radius of @xcite ( see @xcite for details to implement the form factor with the reduced radius ) . as we can see in fig .
[ fig : s0red ] , the binding goes now from about 44 mev when partial omission of charge exchange is considered , eq .
( [ eq : tknneski ] ) , to 38 mev when it is explicitly considered , and then to 35 mev when the @xmath91 channel is explicitly taken into account .
the latter case is our final result , which provides 35 mev binding for this system and a width of about 50 mev . , @xmath62 , and @xmath107 .
the argument is @xmath108 for the three body system .
the argument @xmath109 for the two body system is obtained through eq .
( [ eq : argment ] ) .
the upper x - axis scale is for @xmath110.,scaledwidth=100.0% ] it is interesting to see the origin of the peaks in figs .
[ fig : s0 ] , [ fig : s0red ] . for this purpose
we show in fig .
[ fig : kbarn ] the modulus squared of the @xmath111 , @xmath62 , and @xmath107 amplitudes .
we see that the shape of figs .
[ fig : s0 ] , [ fig : s0red ] is similar to the one of the @xmath111 , which already provides a peak in the total amplitude of eq .
( [ eq : tknncharge ] ) at the single scattering level .
the multiple scattering is responsible for an extra binding of about 25 mev and 10 mev increase in the width .
alternatively , we can view the @xmath18 state as a bound @xmath112 state as suggested in @xcite . in this case fig .
[ fig : kbarn ] could be interpreted as having a non interacting @xmath112 state , where the @xmath113 has been formed in the collision of the @xmath2 with one nucleon and the second one has been a spectator . then , the results of fig . [ fig : s0red ] , where the @xmath2 has reinteracted with the two nucleons , could be interpreted as a system where the @xmath113 has interacted with the second nucleon and a quasibound @xmath112 state is formed @xcite .
we have done a derivation of the fca to the faddeev equations for the case of the @xmath18 system with @xmath1 , complementing what was already done in @xcite for the @xmath6 at threshold , extended below threshold in @xcite . with this new formula
we evaluate the @xmath18 amplitude below the @xmath18 threshold and we find a clear peak of @xmath28 at about 3334 mev , reducing the former binding by about 6 mev , with a width of 50 mev .
this is a moderate change with respect to the one found in @xcite ignoring the charge exchange processes , and also smaller than the effect found in @xcite for the @xmath5 , where the charge exchange processes was found quite important , reducing the binding of the @xmath18 with @xmath5 in about 12 mev .
we have also shown that the explicit consideration of the @xmath3 state in the three body system is rather moderate , and goes in the direction of decreasing the binding by about 3 mev . in summary , here and
in @xcite we have shown that the fca to the faddeev equations is a good tool which provides reasonable values for the binding and width of the @xmath18 system , corroborating qualitatively findings done with other methods which are technically much more involved .
we find a binding for the @xmath18 system with @xmath1 of 35 mev with a width of 50 mev .
this should be compared to the about 20 mev binding in @xcite and 16 mev in a recent paper @xcite using the same input as in @xcite .
even with these differences , the binding obtained here is substantially smaller than other values quoted in the literature which did not use the chiral dynamical approach . as to what to attribute the difference with the results of @xcite , certainly there are approximations done in the fca .
one of them , making a static picture of the two @xmath114 and using eq .
( [ eq : argment ] ) , which relates the @xmath12 effective energy to the total energy , does not consider recoil of the nucleon .
one can make some rough estimate of this effects as follows . since @xmath115 , with @xmath116 the total fourmomentum of the @xmath18 system ( @xmath117 ) , we would have @xmath118 and the first term of eq .
( [ eq : yenis1 ] ) would correspond to the expression of eq .
( [ eq : argment ] ) that neglects recoil .
continuing with the rough approximation we can think that @xmath119 with @xmath120 the binding energy of a single nucleon , where we have assumed the kinetic energy to be of the order of the binding energy and the binding energy of each particle proportional to its mass ( this is also implicit in eq .
( [ eq : argment ] ) ( see @xcite ) ) .
this correction results into a decrease of the binding of the @xmath1 state by about 8 mev , bringing the results closer those of @xcite .
incidentally , the @xmath5 state reported in @xcite would reduce its binding by about 2 - 3 mev .
on the other hand , the variational or faddeev approaches also have intrinsic uncertainties tied to the use of off - shell extrapolation of the amplitudes .
to our knowledge there is only one work where the effects of these off - shell extrapolations has been studied in the three body system .
this is the work of @xcite ( section v ) , which studies the @xmath121 system with faddeev equations , and the off - shell effects were shown to produce a shift of 40 mev in the energy of the @xmath122 resonance ( now called @xmath123 ) .
one can make some guesses on what the effects could be in the present case .
the amount of 40 mev over the excess energy of the @xmath123 over threshold of @xmath121 , 170 mev , is about 24 % .
thus , the same proportion applied to the binding of the @xmath18 system found by us gives 8 mev .
alternatively , we can look at the differences between different approaches using the same input to give us an idea of the possible effects .
inspection of the results of @xcite tells us that the differences are of the order of 10 mev .
thus , effects of the order of 10 mev from unphysical off - shell contributions are likely in the @xmath124 case . on the other hand ,
the @xmath5 case of @xcite could be a privileged one in this sense , since applying the same percentage as before to the 9 mev binding found in @xcite , we find likely off - shell effects of the order of 2 mev .
the findings for the @xmath5 in @xcite , with the uncertainties estimated here could be in agreement with the results for this case in @xcite , where the authors conclude that if there is a quasibound @xmath125 state it must be bound by less than about 11 mev . within the uncertainties of the method , the simplicity of the present approach allows one to get an insight on the problems , following the analytical expressions of the amplitudes .
ultimately , improvements on all these works could be done within the approach of @xcite , which one must encourage , which uses full faddeev equations in coupled channels and relies upon the on - shell t - matrices alone .
we thank a. gal and t. hyodo for useful comments . this work is partly supported by dgicyt contract number fis2011 - 28853-c02 - 01 , the generalitat valenciana in the program prometeo , 2009/090 .
we acknowledge the support of the european community - research infrastructure integrating activity study of strongly interacting matter ( acronym hadronphysics3 , grant agreement n. 283286 ) under the seventh framework programme of eu .
s. s. kamalov , e. oset and a. ramos , nucl .
a * 690 * , 494 ( 2001 ) .
e. oset , d. jido , t. sekihara , a. m. torres , k. p. khemchandani , m. bayar and j. yamagata - sekihara , nucl .
a * 881 * , 127 ( 2012 ) .
m. bayar , j. yamagata - sekihara , e. oset , phys .
rev . * c84 * , 015209 ( 2011 ) .
y. ikeda and t. sato , phys .
c * 76 * , 035203 ( 2007 ) .
n. v. shevchenko , a. gal and j. mares , phys .
lett . * 98 * , 082301 ( 2007 ) .
n. v. shevchenko , a. gal , j. mares and j. revai , phys .
c * 76 * , 044004 ( 2007 ) .
n. kaiser , p. b. siegel and w. weise , nucl .
a * 594 * , 325 ( 1995 ) .
e. oset and a. ramos , nucl .
a * 635 * , 99 ( 1998 ) .
j. a. oller and u. g. meissner , phys .
b * 500 * , 263 ( 2001 ) .
d. jido , j. a. oller , e. oset , a. ramos , u. g. meissner , nucl .
* a725 * , 181 - 200 ( 2003 ) .
l. roca and e. oset , phys .
d * 82 * , 054013 ( 2010 ) .
a. arai , m. oka and s. yasui , prog .
phys . * 119 * , 103 ( 2008 ) .
t. uchino , t. hyodo and m. oka , nucl .
a * 868 - 869 * , 53 ( 2011 ) .
n. barnea , a. gal and e. z. liverts , phys .
b * 712 * ( 2012 ) 132 - 137 . | we extend the fixed center approximation ( fca ) to the faddeev equations for the @xmath0 system with @xmath1 , including the charge exchange mechanisms in the @xmath2 rescattering which have been ignored in former works within the fca .
we obtain similar results to those found before , but the binding is reduced by 6 mev . at the same time we also evaluate the explicit contribution the @xmath3 intermediate state in the three body system and find that it produces and additional small decrease in the binding of about 3 mev .
the system appears bound by about 35 mev and the width omitting two body absorption , is about 50 mev . |
it is nevertheless clear that given a consistent theory ( such as general relativity itself ) its linear part in any analytic expansion should be consistent as well ( up to linear order )
. the object of our concern in the present paper will be the linear deviations from flat minkowski space , i.e. g_=_+h _ where @xmath3 is the minkowski metric , and @xmath4 .
this equation is taken to be an exact one ; it can be looked at as the definition of @xmath1 .
now , it is a fact of life that @xmath5 has got dimensions of length , and that @xmath6 enjoys dimensions of mass .
the value of newton s constant indicates that at the scale of terrestial experiments , @xmath7 .
this means that the field @xmath1 enjoys the proper canonical dimension ( one ) of a four - dimensional gauge field .
the inverse metric is defined as a formal power series : g^=^-h^+^2 h^_h^-^3 h^ h _ h^ + o(^4 ) diffeomorphisms with infinitesimal parameter @xmath8 act on the full metric as g_=()g _ whereas in terms of the fluctuations [ variations ] h_=^_h _ + ( _ _ + _ _ ) + h__^+h__^this is , again , an exact formula , in the sense that there are no @xmath9 corrections to it .
the symmetry as above , without any restrictions , is the one corresponding to general relativity , and in the present paper will be referred to simply as diff . when the vector @xmath10 is restricted to _
^=0 the symmetry is broken to what we call t(ransverse)diff @xcite .
the total action is then defined by s = s_h+s_m here @xmath11 represents the purely gravitational sector , which is the most general lorentz invariant dimension four operator that can be written with the field @xmath12 , and its derivatives .
it can be parametrized by a string of constants , namely the ones associated with the kinetic energy , @xmath13,@xmath14 and the ones associated to the potential energy for the fluctuations , which is the most general quartic potential in the fluctuations @xmath12 , namely , @xmath15 , @xmath16 .
the overall scale of the potential energy is related to the _ cosmological constant _ , @xmath17 .
cosmological observations seem to favor the tiny value @xmath18 .
[ grav ] s_h&&d^4 x ( _ h^^h_- _ h^_h^_+c_2 ^h^h_-
c_3_h ^h + . + & & m_d^4 ( 1 + _1 h + ( _2 h^2 - 2 _3 h_h^)+ .
+ & & .. ( _4 h^3 + 8 _5 h _
h^ h_^- 6 _6 h h_h^ ) + .. + & & .. ( _7 h^4 - 12 _8 h^2 h_h^+32 _9 h h_h^h_^- ... + & & ... 48 _10h_h^ h _
_11(h_h^)^2 ) ) ) let us remark , first of all , that the structure of the symmetry transformations of both diff and tdiff is such that terms in the lagrangian of @xmath19 are related to terns of both @xmath19 as well as , because of the abelian part , terms of @xmath20 .
this means that in the variations of the kinetic energy part we can keep only the piece in @xmath21 , since the other part ( @xmath22 ) of the variation should cancel with the @xmath21 contribution to the kinetic operators of order @xmath23 , which we have not considered .
the only piece we can consistently consider there is then the fierz - pauli abelian part given by h_=m_p(__+ _ _ ) the situation is different , however , in the potential energy piece . in order to cancel the fierz - pauli variation of the @xmath24 term
, it is necessary to consider the @xmath22 variation of the @xmath23 term .
this means that the full action is invariant under the full variation ( [ variations ] ) up to dimension five operators ( which means @xmath23 in the kinetic energy part , and @xmath25 in the potential energy piece ) . under those provisos
, tdiff needs that & & _1=_3=_5=_10 + & & _2=_6=_9=_11 + & & _4=_8 the most general tdiff invariant potential depends on four arbitrary parameters .
in some studies it is frequent to restrict the gravitational equations of motion to the linear approximation ; this means quadratic terms in the gravitational lagrangian . from a field
theoretical viewpoint there is no reason lo leave away any _ relevant _ ( in the renormalization group sense ) operators .
let us first consider the case @xmath26 .
this corresponds to vanishing cosmological constant in general relativity .
first of all , tdiff enforces @xmath27 . besides , there are two exceptional values , namely @xmath28 , when tdiff is enhanced to full diff .
this is the only combination for which the wave operator is transverse .
& & d^h_=-h_+ ( _ _ h_^+__h^_)- ( _ _ _ h^+__h)+ + & & + h _ indeed ^d^h_=^h_+ _ _ _ h^+ _ h by the way , it is worth noticing that the metric condition _
g_=0 is _ identically _ satisfied to @xmath29 and poses no restriction on @xmath1 .
the other remarkable value is @xmath30 where the symmetry is enhanced with a weyl invariance , denoted by wtdiff , and the wave operator is traceless in the absence of cosmological constant . to be specific , ^d^h_=(c_3-)h + ( c_2- )
the analysis in @xcite shows that these two are the only instances where only spin two is present , with no scalar contamination .
let us now consider the effect of @xmath31 .
first of all , diff invariance is recovered when @xmath32 .
curiously enough , as such , and in the quadratic approximation , the term m_1
^ 2_3 as well as m_2 ^ 2_2 do have the interpretation of
_ masses_. only when the background around which we perturb is not flat , but a constant curvature space , with metric @xmath33 , do these paramerers recover the meaning of cosmological constant . in that case
it is mandatory to substitute all derivatives by background covariant derivatives , i.e. , _
h_|_h _ and to raise and lower indices using the corresponding background metric : h^|g^|g^h _
let us study now consistency of the coupling , which was the main motivation of this work .
the matter lagrangian has been denoted by @xmath34 . up to dimension five operators for a scalar field , which in a free falling locally inertial reference system has lagrangian l^0_m=^__- v ( ) assuming a @xmath35 symmetry -the allowed matter operators when a gravitational field is present can be parameterized by three constants , @xmath36 : [ matt ] l_m=^__-v()+ ( - h^__+ _ 2 h^__- _ 3 v ( ) )
remember that the variation of a scalar field is = ^_in order to enjoy tdiff invariance , it is necessary that @xmath37 .
diff invariance needs in addition that @xmath38 .
the matter equations of motion are = -- v^()+ ( _ 1_(h^_)- _ ( h ^)-_3 v^ ( ) ) and the gravitational equations [ graveq ] = d^h_- ( _
1__+ ( _ 3 v()-(_)^2 ) _ ) there is generically no problem of consistency , except in the two exceptional cases .
first of all , when @xmath11 has extended diff symmetry , and consequently a transverse wave operator , this forces @xmath34 to have the same diff symmetry through the linear expansion of the @xmath39 term ; otherwise consistency of the coupling enforces extra condictions on the matter , a very weird situation indeed ( i.e. bianchi identities are still valid on the gravitational side , so that by consistency the same identities must hold true on the matter side as well ) . nevertheless , it is not fully devoid of interest to study the situations in which there are exceptions to this rule ; this we did in a previous work @xcite .
when there is wtdiff symmetry , it is clear that the matter lagrangian should also be scale invariant in order for the corresponding energy - momentum to be traceless . in our example
, this corresponds to @xmath40 and @xmath41 .
there are models in which diff invariance in the matter sector is reached using in the volume element some other scalar density , such as the square root of the determinant of a matrix built out of fields and their derivatives ( as in the very interesting ones proposed in @xcite ) instead of the @xmath39 term implicit in the metric volume element .
the tensor that appears as the source of gravity in einstein s equations is covariantly conserved thanks to the equations of motion of the fields in the scalar density .
of course that tensor is not the usual energy - momentum tensor of general relativity , which now is not conserved .
the reason is that in order for the rosenfeld energy - momentum tensor to be equivalent to the canonical belinfante one what is needed is not only diff invariance , but also the standard metric volume element @xcite .
this topic seems worthy of some further investigation .
in this section we will outline the way in which one can constraint the space of paramaters of the linearized theory ( i.e. @xmath13 , @xmath42 and @xmath15 ) using experimental results on deviations from newton s inverse square law . for simplicity
we will illustrate with a very particular example so no definite conclusions can be drawn concerning the viability of this kind of models .
detailed computations of the propagators can be found in @xcite , where the authors considered a gravitational lagrangian ( [ grav ] ) with all @xmath43 except @xmath44 which has the interpretation of a mass for the scalar part of the graviton , present generically in this kind of models with tdiff invariance .
it turns out that for a conserved energy - momentum tensor coupled to gravity in the form [ coup ] l_i = h_(_1t^+_2^t ) then in momentum space the interaction is l_i=_1 ^ 2-(_2+_1)^2 where we have defined c = c_3-+c_2-c_2 ^ 2 with the constraint @xmath45 because of unitarity @xcite .
the first term corresponds to the usual spin 2 exchange while the second one is an additional massive scalar interaction .
let us turn our attention to a particular example , namely the matter lagrangian ( [ matt ] ) . unfortunately the corresponding energy momentum tensor is not conserved .
however , a conserved tensor can be defined as _ & & t_-_((_)^2+(_3 - 1)v())+ o()= + & = & _ _ -_((_)^2-v())+ o ( ) in the particular case that @xmath46 ( which includes the diff invariant lagrangian ) our energy - momentum tensor can be written in terms of the new one and its trace in such a way that the coupling @xmath47 is of the form ( [ coup ] ) with & & _
1= + & & _ 2= now we can apply directly the preceeding results and study experimental constraints to this model .
the exchange of additional massive scalar degrees of freedom produces a yukawa like potential which is usually parametrized as @xcite v(r)~(1+e^- ) the parameter @xmath48 is then the ratio between the spin 2 and the scalar couplings , in our particular case [ alfa ] = -=- while @xmath49 gives the range of the interaction , or equvalently the mass of the scalar exchanged [ landa ] ^2= notice that one has to impose @xmath50 since as we have said absence of ghosts requires @xmath45 .
there are important constraints on the strength of hypothetical yukawa interactions for a wide range of @xmath49 . through ( [ alfa ] ) and ( [ landa ] )
it is then possible to constraint the space of parameters of the linearized theory .
we will use figures 4 , 5 and 9 of reference @xcite , which show regions allowed and excluded for @xmath48 corresponding to @xmath49 in the ranges @xmath51m-@xmath52 m , @xmath52m-@xmath53 m and @xmath53m-@xmath54 m respectively .
since we are just interested in general behaviours and not in accurate results we will approximate the experimental curves by straight lines .
the original plots are in logarithmic scale so we have experimentally allowed regions of the form [ curve ] ||<k^a we just have to substitute this expresion into ( [ alfa ] ) to get bounds for our parameters .
there are however four parameters to play with ( @xmath55 , @xmath56 , @xmath57 and @xmath58 ) .
first , it is interesting to see the order of magnitude for the mass once we fix the values of @xmath57 and @xmath58 .
the result is ploted in fig.[m2.n2 ] .
it can be seen that greater values for the mass are favoured , being the lower bound around @xmath59 , and that the allowed region rapidly decreases with @xmath60 . ] another possibility is to fix @xmath61 and @xmath55 and see , in the plane ( @xmath57,@xmath58 ) , how far from diff invariance ( which corresponds to @xmath62 ) can we move away
. remember that we also have to take into account the restriction @xmath63 .
for the first range , @xmath64m-@xmath52 m there is no hope of seeing an experimental curve that appreciably deviates from the parabola @xmath65 because of ( [ landa ] ) and the tiny values of @xmath49 .
increasing @xmath61 does increase the allowed region , which is between both curves , but does not produce a plot in which the curves are visibly separate .
it can be understood if we realise that increasing the mass also increases @xmath58 on the parabola through @xmath57 and ( [ alfa])-([landa ] ) .
an approximate definition of the separation could be sep~= where @xmath66 and @xmath67 means the value of @xmath58 on the parabola and the experimental curve respectively .
for the first experimental range one has @xmath68 independently of the mass and in the whole interval .
the other two cases do not have that property , which is of course related also with the particular values of @xmath69 and @xmath70 in ( [ curve ] ) .
examples of resulting plots are figs.[c3.c2.1],@xmath71[c3.c2.2 ] and [ c3.c2.3 ] .
once again the experiment prefers greater values for the mass .
all the plots have @xmath72 , other values just move the experimental curve along the parabola , but the qualitative result remains unchanged . ] ] ]
in this paper we have studied at the linearized level the viability of gravity models with a restricted symmetry , both from the theoretical and observational points of view . while the existing observational constraints on additional yukawa like gravitatory interactions do not seem to be a major obstacle , a consistency problem has been identified . at the non linear level it appeared as an integrability condition on einstein s equations @xcite .
here we turned our attention to the linear level in order to see if the problem could be avoided , and if so in what type of more clever non linear completions .
the main conclusion is that it is not generically possible to couple matter ( i.e. , with an arbitrary equation of state ) to gravitation in such a way that this coupling has a restricted symmetry only ( what has been called tdiff ) whereas the purely gravitational sector enjoys a higher symmetry , namely the standard diff invariance , or an additional weyl symmetry ( wtdiff ) . that this is possible in some restricted cases has been already found in a previous paper @xcite .
the condition for arbitrary tdiff matter to be able to couple to diff gravity without restrictions can be stated somewhat more formally by saying that the rosenfeld ( metric ) energy - momentum tensor has got to be equivalent to the belinfante canonical form .
on the other hand , there is a widespread _ urban legend _ asserting that unimodular theories are equivalent to general relativity with a cosmological constant .
specific calculations both here and in our previous paper @xcite have proven it to be groundless .
it is a fact that in some tdiff models there is no exponential expansion at all which is well known to be the benchmark of a ( positive ) cosmological constant in general relativity .
therefore , that models provide a counterexample to the statement above . nevertheless , as with all legends , there is some partial truth in it .
the equations of motion of the example in @xcite correspond to @xmath73 and @xmath74 , that is [ t ] d^fp_=__whereas the linear equations of general relativity with cosmological constant read : [ gr ] d^fp_+ _ = ( _ _ + ( v()-(_)^2 ) _ ) now , the equations ( [ t ] ) are inconsistent as such , in the sense that only a subsector of the theory , namely , the one that obeys [ ic ] v-(_)^2= c can be coupled to gravitation .
there are many sectors of matter in a freely falling inertial system that do not obey thisrestriction .
actually , together with energy conservation , the aforementioned equation implies that both the kinetic and potential energy ought to be constant : & & 2 v()= e + c + & & ( _ ) ^2 = e - c it is clear that for a scalar field in flat space most initial conditions lead to configurations that violate those equations .
this would mean that an inconsistency would show up once a gravitational field is turned on , however weak .
more formally , something very strange should happen when changing the reference frame from an inertial ( freely falling one ) to another in which a gravitational field is present .
those are the reasons what we say that the coupling is generically inconsistent .
let us accept nevertheless , for the sake of the argument , that physics is so restricted .
then a glance at the equations ( [ gr ] ) of general relativity shows that they are indeed equivalent to ( [ t ] ) _ provided _ we identify but this is only due to our choice of the arbitrary constants , and under the asumption that the coupled sector is only the one that obeys ( [ ic ] ) , a deeply misterious condition from a general relativistic perspective . in the general tdiff case
the analogous condition to ( [ ic ] ) is -(_)^2+(_3 - 1)v()=c and the system is not equivalent to general relativity with a cosmological constant .
a.f.f . would like to thank irene amado for her infinite patience and help with the figures .
this work has been partially supported by the european commission ( hprn - ct-200 - 00148 ) and by fpa2003 - 04597 ( dgi del mcyt , spain ) , as well as proyecto hephacos ( cam ) ; p - esp-00346 .
a.f.f has been supported by a mec grant , ap-2004 - 0921 .
e. alvarez , d. blas , j. garriga and e. verdaguer , `` transverse fierz - pauli symmetry , '' nucl .
b * 756 * ( 2006 ) 148 [ arxiv : hep - th/0606019 ] .
+ e. alvarez , `` can one tell einstein s unimodular theory from einstein s general relativity ?
, '' jhep * 0503 * ( 2005 ) 002 [ arxiv : hep - th/0501146 ] .
s. v. babak and l. p. grishchuk , `` the energy - momentum tensor for the gravitational field , '' phys .
d * 61 * ( 2000 ) 024038 [ arxiv : gr - qc/9907027 ] .
d. blas , `` gauge symmetry and consistent spin - two theories , '' j. phys . a * 40 * , 6965 ( 2007 ) [ arxiv : hep - th/0701049 ] . | we point out a generic inconsistency of the coupling of ordinary gravity as described by general relativity with matter invariant only under unimodular diffeomorphisms ( tdiffs ) , and some previously studied exceptions are pointed out .
the most general lagrangian invariant under tdiff up to dimension five operators is determined , and consistency with existing observations is studied in some cases .
ift - uam / csic-07 - 32 + arxiv : yymm.nnnnvv + * a comment on the matter - graviton coupling * + 0.4 cm _ instituto de fsica terica uam / csic , c - xvi , and departamento de fsica terica , c - xi , + universidad autnoma de madrid e-28049-madrid , spain _ 0.2 cm |
a _ deterministic finite automata _ ( dfa ) @xmath8 is a triple @xmath9 where @xmath10 is the state set , @xmath11 is the input alphabet and @xmath12 is the _ transition function_. if @xmath13 is completely defined on @xmath14 then @xmath8 is called _ complete _ , otherwise @xmath8 is called _
partial_. the function @xmath13 extends uniquely to a function @xmath15 , where @xmath16 stands for the free monoid over @xmath11 ; the latter function is still denoted by @xmath13 .
when we have specified a dfa @xmath17 , we can simplify the notation by writing @xmath18 instead of @xmath19 for a subset @xmath20 and a word @xmath21 . in what follows
, we assume @xmath22 because the singleton alphabet case is trivial for considered problems .
given a subset of words @xmath23 and a state @xmath24 denote by @xmath25 the subset of words from @xmath26 labeling some path from @xmath24 in @xmath8 . a dfa @xmath27 is called _ synchronizing _ if there exists a word @xmath28 such that @xmath29 .
notice that here @xmath30 is not assumed to be defined at all states .
each word @xmath30 with this property is said to be a _
word for @xmath8 .
the minimum length of such words is called the _ reset threshold _ of @xmath8 and is denoted by @xmath31 .
the synchronization of strongly connected partial automata as models of _ @xmath0-machines _ is one of the central object for research in the theory of stationary information sources .
the synchronization and state prediction for stationary information sources has many applications in information theory and dynamical systems .
an @xmath0-machine can be defined as a strongly connected dfa with a probability distribution defined on the outgoing arrows for each state , without states with _ probabilistically equivalent future _ ( see @xcite,@xcite for details ) .
hence by @xmath0-machine we mean the tuple @xmath32 where @xmath33 is the strongly connected partial automaton and @xmath34 for the probability of generating @xmath35 from @xmath36 instead of @xmath37 is the probability distribution on the outgoing arrows .
formally , * @xmath38 for each state @xmath39 ; * @xmath40 whenever @xmath35 is undefined on @xmath39 ; * for each states @xmath41 there is a word @xmath42 such that @xmath43 , where the probability @xmath44 of generating a given word @xmath45 from the state @xmath24 is defined inductively by @xmath46 . an @xmath0-machine is _ exactly synchronizable _ or simply _ exact _ if the corresponding partial strongly connected automaton is synchronizing in our terms .
given an initial probability distribution @xmath47 on the states of an @xmath0-machine , the probability of generating a word @xmath48 is equal to the weighted sum @xmath49 .
based on various applications , there are two basic settings for synchronization . in the first
setting , the observer knows that an error appeared in the corresponding system and he can apply the reset sequence for reestablishing correct behavior of the system .
apparently , the most natural measure of synchronization in this setting is the reset threshold of the corresponding automaton .
this is because the reset threshold corresponds to the minimum time required to find out the current system state . in the second
setting , observer either does nt know that an error appeared or can not affect to the system after that . for such situations
it is natural to estimate either the probability of generating a reset word ( if it exists ) by the system itself or the average uncertainty in predicting the current state by the generated sequence in a given period of time .
let @xmath50 be an exact @xmath0-machine and denote @xmath51 .
it has been proved in @xcite that the probability of generating only non - reset words of length @xmath1 decay exponentially fast as @xmath52 where @xmath53 is a so - called _ synchronization rate constant _ can be defined as @xmath54 where @xmath55 is the set of non - reset words of length @xmath1 and @xmath56 is a steady state distribution on the states .
since @xmath8 is strongly connected , this value does nt depend on the initial distribution @xmath56 .
it is also shown in @xcite that the synchronization rate constant can be approximated with any given accuracy in exponential time in terms of the number of machine states @xmath57 . in this section
we show how to approximate @xmath58 with any given accuracy in polynomial time . as well as in @xcite
, for this purpose , let us consider the auxiliary _ _ semi @xmath0-machine__-machine we mean a partial automaton with edges weighted by real numbers in the range @xmath59 $ ] . we still refer to these weights as probabilities . ]
@xmath60 where @xmath61 . given a pair of different states @xmath62 and a letter @xmath63 ,
define @xmath64 @xmath65 let @xmath66 be the @xmath67 matrix of the transition probabilities of @xmath68 for the letter @xmath69 indexed by the states of @xmath70 .
that is , for each two states of @xmath71 the entry on the intersection of @xmath72-th row and @xmath73-th column is given by @xmath74 the transition probability matrix @xmath75 of @xmath68 is now defined as the sum @xmath76 let @xmath77 be the sum of @xmath78-th row entries of @xmath79 .
notice that by definition [ pr2def ] , @xmath77 is the probability of generating words by @xmath0-machine @xmath8 of length @xmath1 from @xmath24 which does nt merge the pair @xmath80 .
define also @xmath81 a construction of a polynomial time algorithm for approximation synchronization rate constants is based on the following theorem .
[ th_comp_synchconst]for each state @xmath82 the probability of generating a non - reset sequence of length @xmath1 from @xmath24 is upper bounded by @xmath83 and is lower bounded by @xmath84 .
let @xmath48 be a non - reset word of length @xmath1 generated by @xmath8 from the state @xmath24 .
then there exists another state @xmath39 such that @xmath48 does nt merge @xmath24 and @xmath36 in @xmath8 .
this means that @xmath48 takes the pair @xmath78 to another proper pair @xmath85 and thus the probability of generating this word is included in @xmath77 .
the upper bound follows .
now let @xmath36 be the state such that @xmath86 .
let @xmath42 be a word of length @xmath1 which does nt merge the pair @xmath87 , that is , @xmath42 is an arbitrary word which probability is included in the @xmath78-th row .
then @xmath42 is non - reset and the lower bound follows .
since @xmath8 is strongly connected and synchronizing , it has a unique positive steady state distribution @xmath88 ( see e.g. @xcite ) .
hence we get the following corollary .
[ cor_bounds ] @xmath89 thus we get that the probability of generating non - reset words of length @xmath1 is lower bounded by @xmath90 and is upper bounded by @xmath91 .
taking the power @xmath92 we get that the synchronization rate constant is given by the maximal eigenvalue of the transition probabilitiy matrix @xmath75 whence the main result of this section follows .
[ main_th ] given an exact @xmath57-state @xmath0-machine @xmath8 , its synchronization rate constant can be approximated in a polynomial time @xmath93 for arbitrary small absolute error @xmath94 .
notice also , that corollary [ cor_bounds ] can be used to estimate the probability of generating non - reset words of a given length in a polynomial time .
in the previous section we show how to compute synchronization rate constants for exact @xmath0-machines .
it turns out that if an @xmath0-machine is not exact , it is still can be synchronized but only asymptotically , that is , for almost every infinite word @xmath95 the observer uncertainty decay exponentially fast as @xmath96 after reading @xmath1 first symbols of @xmath97 @xcite .
let an @xmath0-machine @xmath98 generates a word @xmath30 , denote by @xmath99 the observer _ belief distribution _ , that is , @xmath100 where @xmath101 and @xmath56 is the initial distribution of @xmath8 . clearly , @xmath102 is a stochastic vector and @xmath36-th entry is equal to the probability that @xmath8 is in the state @xmath36 after generating @xmath30 . in particular , if @xmath30 is reset then @xmath102 has only one non - null entry , equals @xmath4 .
let @xmath103 be the random variable for the belief distribution over states induced by the first length-@xmath1 word the machine generates , and @xmath104 be the most likely state in @xmath105 ( if a tie the lowest numbered state is taken ) .
denote by @xmath106 the combined probability of all other states in the distribution @xmath105 .
then @xmath8 is said to be _ asymptotically synchronizing _ if @xmath107 vanishes to @xmath108 while @xmath109 for each @xmath94 .
in other words , for almost every infinite word the observer uncertainty in predicting the current machine state vanishes to @xmath108 ( see @xcite for details ) .
it is proved in @xcite that actually each @xmath0-machine is asymptotically synchronizing , and @xmath110 vanishes to @xmath108 exponentially fast , that is , @xmath111 also vanishes to @xmath108 for some @xmath112 while @xmath113 . however , no algorithm for computing infimum of such @xmath35 is given in @xcite .
let us call this infimum the _ prediction rate constant _ of an @xmath0-machine and denote it by @xmath114 .
below we show that the prediction rate constant is always positive whenever @xmath0-machine is not exact , present a polynomial time algorithm for approximating it , and simultaneously prove the aforementioned result from @xcite , apparently in a simpler way .
notice also that the prediction rate constant is equal to @xmath108 for exact @xmath0-machines . in
what follows , let @xmath8 be non - exact .
suppose @xmath8 generates a word @xmath30 .
denote by @xmath115 the state with maximal @xmath116 among @xmath82 and by @xmath117 the state with maximal @xmath118 among @xmath82 such that @xmath119 ( if a tie the lowest numbered state is taken ) . in these terms
, we can bound @xmath120 as follows .
[ lem_ql_bounds ] for some constants @xmath121 depended only on the machine , we have @xmath122 by the definition of @xmath110 we have @xmath123 for @xmath39 with maximal @xmath124 . since @xmath125 , either @xmath126 or @xmath127 whence at least one of the components @xmath128 is contained in the sum @xmath129 . since also @xmath130 we get @xmath131 where @xmath132
by the choice of @xmath36 and @xmath117 we get @xmath133 by the definition of @xmath115 we also get @xmath134 thus we can choose @xmath135 and the lemma follows .
the following lemma gives us a way to calculate the prediction rate constant .
[ lem_calc_pr ] @xmath136 let @xmath137 be some constants . if @xmath138 does not vanish to @xmath108 , then @xmath139 for @xmath1 big enough . from the other hand , if @xmath140 vanishes to @xmath108 then @xmath141 for @xmath1 big enough .
the lemma follows .
thus , in order to design an algorithm for computing prediction rate constant , it is enough to calculate @xmath142 and then to prove that @xmath143 for these purposes we need the definition of associated _ edge - machine _ as in @xcite .
let @xmath144 be an ergodic irreducible markov chain with the equilibrium distribution @xmath145 .
let @xmath146 denotes the indicator function of the transition from the state @xmath147 to the state @xmath148 by the word @xmath30 .
that is , @xmath149 if @xmath150 and @xmath108 otherwise .
the edge machine @xmath151 is the markov chain whose states are the outgoing edges of the original machine @xmath144 .
that is , the states are the pairs @xmath152 such that @xmath153 , and the transition probabilities are defined as : @xmath154 .
a sequence of @xmath151 states visited by the markov chain corresponds to a sequence of edges visited by the original machine @xmath144 with the same probabilities .
it follows , that @xmath151 is also ergodic and the following remark is also straightforward .
[ rem_edge_prob]the equilibrium distribution of @xmath151 is given by @xmath155 where @xmath156 .
in contrast to @xcite , we consider @xmath151 machines for _ deadlock components _ of original semi - machine @xmath68 instead of @xmath8 . as in the previous section we consider the same auxiliary automaton @xmath157 . since @xmath8 is not exact , there are pairs @xmath80 that can not be synchronized , that is @xmath158 for each word @xmath42 ( this follows from @xcite[theorem 3 ] ) .
such pairs are called _
deadlock_. let @xmath80 be a deadlock pair .
it follows from the definition , that for each @xmath159 if @xmath160 is defined then @xmath161 is also defined and @xmath162 is also deadlock .
hence there are closed under the actions of the letters strongly connected components of deadlock pairs in @xmath68 .
since @xmath8 is aperiodic , these components are also irreducible @xmath0-machines .
let us denote the set of such components by @xmath163 .
the following remark follows from the fact that the probability of coming to closed components of the graph is positive .
[ rem_out_of_m]there is a constant @xmath164 such that @xmath165 in simpler terms , it is stated in the remark that with probability asymptotically closed to @xmath4 all pairs of states in @xmath68 either merge or come to deadlock pairs .
let us also notice that the constant @xmath166 from the remark is given by the dominant eigenvalue of the transition matrix of @xmath68 without deadlock components whence it can be approximated in polynomial time . as in
@xcite we use the theorem from @xcite which in simpler terms can be formulated as follows .
[ lem_mc_vars]let @xmath167 be a finite - state , irreducible markov chain , with state set @xmath168 and equilibrium distribution @xmath169 .
let @xmath170 , and @xmath171 .
denote also @xmath172 .
then , there exist @xmath173 such that for each @xmath174 and each state @xmath147 for @xmath1 big enough we have @xmath175 now let @xmath176 and @xmath151 be the corresponding edge - machine .
given a state @xmath177 of @xmath151 , or equivalently an edge between two states @xmath78 and @xmath85 labeled by @xmath97 , define @xmath178 .
due to lemma [ lem_mc_vars ] for any @xmath174 , we have @xmath179 for @xmath1 big enough . given any initial pair of states @xmath78 and a random word @xmath30 of length @xmath1 , we have @xmath180(w ) = \frac{1}{l}\ln{\frac{\textbf{p}_{p}(w)}{\textbf{p}_{q}(w)}}\ ] ] by the definition of @xmath151 and @xmath181 .
hence we get @xmath182 where @xmath183 .
[ lem_about_avg]@xmath184 for each @xmath185 .
it is enough to prove that for each pair @xmath78 of @xmath144 the sum @xmath186 is non - negative and there is a pair for which this sum is positive .
since @xmath187 , by remark [ rem_edge_prob ] this sum equals @xmath188 let us use lagrange s method to prove the inequalities .
consider the function @xmath189 by taking derivatives for each variable we get that @xmath190 can have minimum only in the solution points of the system @xmath191 and in the boundary points @xmath192 ( since @xmath193 for @xmath194 ) .
if for some @xmath195 then for each @xmath196 other @xmath197 and @xmath198 . in the opposite case , we have @xmath199 for each @xmath200 .
since @xmath201 this can happen only if @xmath202 for each @xmath203 whence @xmath204 .
thus @xmath204 if and only if @xmath202 for each @xmath205 such that @xmath206 and @xmath207 otherwise . due to the condition of states non - equivalence
we get that for some pair @xmath78 in @xmath144 this condition does nt hold .
the lemma follows .
[ lem_lower_bound]for each @xmath176 and each @xmath94 we have @xmath208 for @xmath1 big enough . first notice
that for each pair of states @xmath209 we get @xmath210 this follows from the fact that one can assign to each word @xmath211 generated from @xmath212 the unique word @xmath213 generated from @xmath214 , where @xmath215 is a fixed word labeling some path from @xmath214 to @xmath212 .
this means that we can take any state @xmath24 and consider only words generated from @xmath24 to estimate the limit of @xmath216 .
given a word @xmath30 of length @xmath1 generated from @xmath24 by lemma [ lem_ql_bounds ] we get @xmath217 if @xmath218 then @xmath219 .
in the opposite case , @xmath220 for any @xmath221 . hence summing up for all length @xmath1 words we obtain @xmath222 where @xmath223 is some state such that the pair @xmath224 belongs to @xmath144 .
suppose @xmath78 does nt belong to any @xmath176 . by remark [ rem_out_of_m ] and inequality [ eq_aep ] for both summands and each @xmath174 for @xmath1 big enough we have @xmath225 where @xmath226 contains the pair
@xmath85 , @xmath227 = ( p',q')}{\textbf{p}_{p}(w ) } \geq 0 , \sum{\lambda_{p',q ' } } + \lambda_p = 1,\ ] ] @xmath228 is from inequality [ eq_aep ] and @xmath229 is the maximum of @xmath230 among all @xmath63 and @xmath39 such that @xmath231 . since @xmath232 and for each @xmath0 this inequality holds for @xmath1 big enough , the lemma follows .
[ lem_upper_bound ] for @xmath176 with minimal @xmath233 and each @xmath234 we have @xmath235 for @xmath1 big enough . by lemma [ lem_ql_bounds ]
we have @xmath236 . using remark [ rem_out_of_m ]
we also get that @xmath237 where @xmath238 .
thus by inequality [ eq_aep ] for each @xmath174 we get @xmath239 for @xmath1 big enough .
the lemma follows .
now our second main result follows from lemma [ lem_lower_bound ] and lemma [ lem_upper_bound ] .
[ th_main_pred]let @xmath8 be a non - exact @xmath0-machine .
then the prediction rate of @xmath8 is given by the maximum of @xmath240 among @xmath185 .
therefore since @xmath241 by lemma [ lem_about_avg ] @xmath242 and @xmath8 is asymptotically synchronizable .
the following corollary follows from the fact that the dominant eigenvectors of @xmath243 can be approximated in polynomial time . given a non - exact @xmath0-machine @xmath8 ,
the prediction rate of @xmath8 can be approximated in polynomial time within any given precision .
thus it turns out that there are polynomial - time approximation schemes ( ptas ) for computing the natural measures of synchronization for stochastic setting for both exact and asymptotic cases while no polynomial time algorithm can approximate the reset threshold even with logarithmic performance ratio , and even for the binary alphabet case @xcite ( see also @xcite,@xcite,@xcite ) unless @xmath244 . | it is known , that an @xmath0-machine is either exactly or asymptotically synchronizing . in the exact case
, the observer can infer the current machine state after observing @xmath1 generated symbols with probability @xmath2 where @xmath3 is a so - called synchronization rate constant . in the asymptotic case
, the probability of the correct prediction the current machine state after observing @xmath1 generated symbols tends to @xmath4 exponentially fast as @xmath5 for @xmath6 and the infimum of such @xmath7 is a so - called prediction rate constant .
hence the synchronization and prediction rate constants serve as natural measures of synchronization for @xmath0-machines . in the present work
we show how to approximate these constants in polynomial time in terms of the number of machine states . |
following the introduction of unification schemes for active galactic nuclei ( agn ) , it is now widely accepted that the central nucleus of an active galaxy has a kerr black hole , a relativistic accretion disk , jets , a dusty torus and clouds emitting broad emission lines .
jets and accretion disks are strongly related through a symbiosis , and we postulate that there may also be a large scale symbiosis involving dust components . since @xmath0-rays are attenuated by photon - photon pair production collisions with lower energy photons from the disk , blr , or torus , such a symbiosis could have an effect on the production and absorption of @xmath0-rays in agn as the blr and torus are powered by the disk , and so may depend on the way in which the jets are fed by the accretion disk .
a @xmath0-ray flare could occur , for example , when a significant amount of energy has been accumulated in the inner part of the accretion disk and is subsequently expelled into the jets . as a consequence
, relativistic shocks may form in the jets and could account for the observed rapid and high amplitude @xmath0-ray variability .
the intense emission observed during a flare could be also due to an increase in the accretion rate , in the bulk lorentz factor of the emitting material in the jet , or in the efficiency of particle acceleration .
the production of @xmath0-rays from relativistic jets depends also on the density of radiation fields since these photons can be scattered by relativistic electrons to gev and tev energies , the external radiation field giving additional target photons to the internal synchrotron radiation assumed in the synchrotron self - compton ( ssc ) process @xcite .
mannheim @xcite and protheroe @xcite have shown that the disk radiation can be relevant for @xmath0-ray production if the main constituents of the jet are relativistic protons . in this paper
we shall concentrate on the role played by external radiation fields in the absorption of @xmath0-rays by photon - photon pair production , and we shall calculate the pair production opacity for gev
tev @xmath0-rays emitted in the jets of quasars and blazars .
we will only consider external radiation fields existing in agn , and will not discuss pair production by @xmath0-rays interacting with internal radiation fields in the emission region as this has been discussed elsewhere @xcite .
we shall also not address the problem of absorption by the infrared background radiation during propagation to earth ( e.g. refs .
@xcite ) .
we shall assume that the disk radiation , the blr radiation , and the torus radiation is anisotropic , but is symmetric about the jet axis .
in such a radiation field the reciprocal of the mean interaction length for photon - photon pair production by @xmath0-rays traveling along the jet axis is given by @xmath1 where @xmath2 is the angle between the direction of the soft photons and the jet axis , @xmath3 is the photon - photon cross section for pair production at center of momentum frame energy squared @xmath4 , @xmath5 is the differential photon number density , and @xmath6 and @xmath7 are defined by the disk , blr , or torus geometry involved or , in the case of @xmath7 , by the pair production threshold condition .
we calculate the optical depth for absorption of gev and tev photons from the point of @xmath0-ray emission to infinity , in the relevant external radiation fields . in quasars
the torus and the blr are active components , as can be inferred from the large forest of emission lines observed , and from the thermal infrared spectrum , allowing good estimation of the diffuse radiation within the central parsec .
in contrast , blazars , which comprise bl lacs and optically violently variable ( ovv ) quasars , are a special category of agn having the jet aligned closely to the line of sight .
the lack of emission lines and uv bump in blazars is usually assumed to be due to a low thermal activity at the centre , implying that the energy density of the external radiation is small in blazars compared to that in quasars
. we shall investigate whether or not this is the case , and discuss how the @xmath0-ray output could be different in quasars and blazars depending on the dominant radiation field .
in section 2 we shall discuss the optical / uv emission from an accretion disk in symbiosis with the jet , and its effect on the blr emission clouds ( ionized by disk photons ) and the torus ( heating of dust ) .
it is difficult to estimate the disk luminosity in blazars as has been done for some quasars with visible uv bumps @xcite .
blazars do not show thermal disk emission , and their spectra from radio to uv frequencies , sometimes to x - ray frequencies , are universally attributed to synchrotron emission from the relativistic jet .
the nonthermal boosted radiation from the jet dwarfs any thermal emission , making the direct detection of blr or torus activity , and hence direct measurements of external photon fields present within the parsec scale of blazars , impossible .
we assume that the accretion disk surrounding the black hole and the relativistic jets are related by a small - scale symbiosis @xcite , and that the energy available for the jet is the energy which would be dissipated between the last marginally stable circular orbit ( in kerr geometry ) , @xmath8 where @xmath9 is the gravitational radius , and the radius of the base of the jet @xmath10 .
the total power of the jet depends on @xmath11 .
a thick jet base means that more energy is expelled into the jet , and therefore , less energy is dissipated in the disk .
a direct consequence of this assumption is that an accretion disk in symbiosis with jets ( adj ) produces a disk luminosity smaller than that of a `` standard '' @xcite relativistic accretion disk .
we suggest that the adj model could describe low - luminosity accretion disks in some blazars , and may provide a bridge between the standard accretion disk assumed to exist in quasars and fanaroff - riley class ii ( fr - ii ) sources @xcite and the low - power advection dominated accretion disk which may be found in some fr - i sources . in section 3 , assuming reasonable models for the size and distribution of blr clouds , we calculate the energy density of the radiation field of the blr , and the optical depth for absorption of gev and tev photons .
broad line regions surround accretion disks and may also be important sources of soft photons which could be inverse - compton scattered by relativistic electrons in the jet to produce @xmath0-rays in `` external compton models '' @xcite . in section 4 , we discuss the infrared radiation from a dusty torus .
based on the assertion that fr - i and bl lacs are related @xcite , and that there is strong evidence that some fr - i have dusty tori @xcite , we assume that blazars must also have dusty tori @xcite and discuss what evidence there is to support this .
since the infrared emission produced by tori could inhibit the escape of tev @xmath0-rays due to photon - photon pair - production @xcite we study how changes in disk activity can modify the geometry and the heating mechanism of the torus , and how this affects the optical depth for @xmath0-@xmath0 absorption in the torus infrared radiation .
relativistic outflows are very nicely described by a model of the small - scale symbiosis between the accretion disk and jets @xcite .
donea and biermann @xcite have used such a model for an accretion disk with jets starting at the inner region of a disk , and were able to reproduce the uv bump in quasars . by fitting observed spectra they derived upper limits for the radius of the base of the jet , @xmath11 , and they found that @xmath11 can not be too far from @xmath12 .
the jet base , or `` footring '' , is actually the thin layer between @xmath12 and radius @xmath11 , the jet being approximated by a hollow cylinder of inner radius @xmath12 and outer radius @xmath11 , such that @xmath13 , where @xmath14 is the outer radius of the disk .
the adj model assumes that the gravitational potential energy available between @xmath12 and @xmath11 is the energy reservoir of the jets .
the total power of the jets is strongly dependent on the accretion mass rate in the disk and on the size of its footring ( we not include here the interaction between the black hole and jets which could also result in it putting a non - negligible fraction of its energy into the jets @xcite ) . the total power of the jets is @xmath15 where @xmath16 is the luminosity of the disk with jets and @xmath17 would be the luminosity of the disk if the conditions necessary to drive outflows were not met , i.e. , the case of a standard relativistic disk .
a coupled jet - disk system must obey the laws of conservation of mass and angular momentum .
we assume that the jet is fed with mass by the accretion disk , and that the flow of mass into the jet per unit of time @xmath18 is a fraction @xmath19 of the accretion mass rate into the disk @xmath20 , where @xmath21 is the surface density of mass in the disk and @xmath22 is the radial velocity of gas at given radius @xmath23 .
the equation for conservation of mass requires @xmath24 .
derivation of the relevant equations can be found in donea and biermann @xcite , who followed the standard method of calculating the emission spectrum from an accretion disk @xcite .
if there is no angular momentum and mass loss into the jet the equations used for mass and angular momentum transport in the disk become the standard equations @xcite , with @xmath25 , @xmath26 and @xmath27 . in this paper
, we shall adopt @xmath28 as used by falcke et al .
@xcite in interpreting the radio - uv correlation in agn . the local physics at the inner radius of the disk ( radius @xmath29 )
is directly related to the extraction of angular momentum from the in - falling gas , and so modifies the structure of the relativistic disk @xcite . from this one
calculates the dissipation energy at radius @xmath23 , @xmath30 , which must be done numerically in the case of a kerr black hole , and a detailed discussion of this would serve no useful purpose here .
instead , for the purpose of illustration , and for the sake of simplicity we give here only the relation for the simpler case of a schwarzschild black hole : @xmath31\ ] ] where @xmath32 and @xmath33 is the black hole mass ; nevertheless , in the present paper we use @xmath30 for the disk of a kerr black hole .
a reasonable outer radius of the footring of the jet would be @xmath34 pc where @xmath35 .
the disk / jet symbiosis is reflected mainly in a modified photon spectrum from the inner region of the accretion disk where the jet is anchored .
the important result is that the spectrum from a kerr accretion disk is cut off at high frequencies , from extreme uv to soft x - rays . in fig .
[ fig1 ] we plot disk luminosities for the adj model , versus the thickness of the footring of the jet for kerr black holes with masses @xmath36 and @xmath37 , and for different mass accretion rates given by @xmath38 , where @xmath39 is the eddington accretion rate .
the case @xmath40 corresponds to an accretion disk without jets ( `` standard '' accretion disk ) and @xmath41 .
as can be seen , for thicker jet bases more energy is available to power jets ( dotted curves show @xmath42 increasing with @xmath43 ) and less energy is radiated by the disk ( solid curves show @xmath44 versus thickness ) .
we note that a small variation of the geometry at the coupling between jet and disk at radii @xmath11 less than @xmath45 would induce large variations in the power of the jet , possibly causing flare activity in blazars .
this is because , in a kerr metric , at small radii close to the black hole there is a large amount of gravitational potential energy available for dissipation into the jet .
because of the jet / disk coupling , the adj model predicts a lower disk luminosity and a softer photon spectrum , and some of the implications of this for interactions of accelerated electrons or protons are discussed in a separate paper @xcite . as mentioned earlier
, a flaring state could arise from enhanced agn central activity whereby the jets get more energy from the disk . in this case , the disk changes from an adj with weak jets into an adj with a spectrum cut off at high frequencies but with correspondingly stronger jets .
since , in this case , the average energy of photons from an adj with strong jets ( flaring ) becomes lower than the average energy of photons from an adj with weak jets this will change the shape of the resulting @xmath0-ray spectrum in models @xcite where disk photons are inverse - compton scattered by relativistic electrons in the jet .
luminosities of disks in blazars are probably @xmath46 erg / s , while quasars typically have @xmath47@xmath48 erg / s .
the adj model gives a simplified approach to the symbiosis between the disk and jets , and could describe low - luminosity accretion disks for quasars and some blazars .
however , for some blazars the adj model could still give uv fluxes much higher than those observed , and in this case an ejection dominated accretion flow ( edaf ) model , which is an adj model with a wind @xcite , may be better .
an edaf model has been applied to the central activity of sgr a * where it was shown that a wind plus jets extracts energy from the disk more efficiency .
since the edaf model does not leave much energy to be dissipated in the accretion disk , the radiation field from the disk would become even less important for gev and tev @xmath0-ray absorption .
alternatively , the jets could extract more energy from the accreting gas if the disk with a jets turns into an advection dominated accretion flow ( adaf ) @xcite .
[ fig10](a ) shows the optical depths from @xmath49 pc to infinity for absorption of gev - tev photons produced in the jet and interacting with photons from an adj with @xmath28 .
we see that at 1 tev photons are only absorbed for high black hole masses and mass accretion rates close to the eddington limit .
this could apply to quasars where the disk radiation field has a high density around the base of the jet . for low accretion mass rates with @xmath50 , and lower black hole masses , @xmath51 ,
the photon - photon optical depth is @xmath52 at 1 tev .
we note that the disk luminosity corresponding to the case @xmath53 for @xmath54 lies roughly at the boundary between blazars and quasars . in fig . [
fig10](b ) we plot the optical depth divided by @xmath55 erg s@xmath56 , and give the likely range for blazars and quasars using @xmath53 for @xmath54 as the boundary . in blazars
we may have adj models with accretion rates as low as @xmath570.001 at lower accretion rates an adaf or edaf model including jets may be more realistic .
we use @xmath54 and @xmath57 as the low luminosity boundary of blazars with an adj in fig .
[ fig10](b ) .
similarly , we use a supermassive black hole ( @xmath58 ) and accretion close to the eddington limit ( @xmath53 ) for the upper luminosity boundary of quasars in fig .
[ fig10](b ) . of course
, these boundaries are by no means rigid and there will be some overlap between the two populations , particularly when we plot @xmath59 . the adaf model seems to explain better very low luminosity blazars , but it can not energize the blr or heat efficiently dust in the torus ; the lack of observed blr and torus activity for some blazars has been sometimes interpreted as suggesting that blazars do not have tori or blr .
we consider that if blazars are to be included in the unification schemes they should also have blr and dusty tori , but possibly with lower emission than in quasars , and we shall discuss this possibility later .
we are interested in obtaining the optical depth for absorption of gev and tev photons in the anisotropic radiation field of the blr . for this
we need to know the distribution of the blr clouds around the jet axis .
broad line regions in agn are not spatially resolved , but information about their size and the gas distribution can be obtained by reverberation mapping , from which it appears that blr are stratified , and have typical sizes of the order of light - weeks in seyfert galaxies and light - months in quasars @xcite .
despite the fact that the majority of bl lacs do not show any emission lines , there are observations suggesting that there is a weak broad @xmath60 emission line in bl lacertae itself @xcite , and even mrk 421 seems to have a very low blr luminosity , @xmath61 erg / s @xcite .
interestingly , celotti et al .
@xcite have suggested , based on the rapid tev variability of mrk 421 , that its accretion rate is lower than @xmath62 of the eddington rate .
hence , a less efficient adj or an adaf may then power the blr for this source .
for mrk 501 , the blr luminosity has been found to be @xmath63 erg / s @xcite .
in addition , chiaberge et al .
@xcite have concluded that blr and obscuring tori are closely linked in quasars , and both are present only in association with accretion . in order to model the blr , and hence its radiation field , we need to know its size and the emission line emissivity as a function of radius .
the two - phase ( hot / cold ) model of the blr predicts a relatively thin shell - like blr @xcite .
higher ionization lines are found at smaller radii ( e.g. for ngc 5548 @xcite ) , and this may favour a thin shell model for the blr .
this provides a very simple model of the blr in which the total blr region luminosity @xmath64 is emitted from the surface of a sphere of radius @xmath65 , i.e. @xmath66 per unit area .
although it is not known if strong winds can exist in blazars , there are models which explain the formation of clouds from bloated stars exposed to winds from accretion disks @xcite , and shocks from jets could also contribute to cloud formation @xcite . on the other hand , frometh and melia @xcite
suggest that the unsteady , turbulent accretion flow onto the black hole is subjected to several disturbances capable of producing shocks , which would heat the gas and allow for formation of blr clouds .
they also suggest that circular clouds originate at smaller radii .
results from reverberation mapping @xcite suggest that the blr in seyferts is extended with an outer radius at least 10 times larger than the inner radius , and kaspi and netzer @xcite use models with a radial dependence of blr cloud number density and radius to fit their data .
clearly , there are several possible mechanisms for formation of blr clouds and it is not known which one operates in blazars
. it could be that all these mechanisms play a role in blr cloud formation .
we shall adopt two extreme cases for the distribution of blr cloud emissivity : ( i ) a geometrically thin shell at radius @xmath65 , ( ii ) a geometrically thick shell extending from radius @xmath67 to @xmath68 with the radial dependence of line emissivity based on the best fit parameters of kaspi and netzer @xcite used in their modelling of the blr in ngc 5548 ( number density of clouds @xmath69 , cloud radius @xmath70 ) . in both cases
we shall assume that the line emission is isotropic , and that the ionization source ( the accretion disk ) is located at @xmath71 and has luminosity @xmath72 .
accretion disk photons excite the blr material , and emission lines such as @xmath60 , @xmath73 , @xmath74 , @xmath75 , etc .
, are produced .
we simplify our problem by making the approximation that the entire blr luminosity is emitted in the @xmath60 ( 6563 ) line .
the blr is assumed to be optically thin with optical depth @xmath76 such that the total blr luminosity is @xmath77 .
celotti et al.@xcite analyzed a sample of quasars and found an average value of @xmath78 erg / s , and for a sample of 12 blazars ( including mrk 421 and mkr 501 ) found an average value of @xmath79 erg / s .
however , flat spectrum radio quasars and ovv quasars could have higher @xmath64 as required by blazejowski et al .
@xcite in order to explain the @xmath0-ray spectra for these objects . in our calculations we use @xmath80 erg / s and @xmath81 , which are typical values used for fitting the broad emission lines with photoionization models ( for @xmath82 , @xmath83 erg / s ) . for @xmath0-ray attenuation by photon - photon pair production
we need the photon number density per solid angle of blr photons along the jet axis , @xmath84 where @xmath85 is the intensity .
calculation of the intensity is straightforward for distribution ( i ) where the blr is a geometrically thin shell . in fig .
[ fig5 ] we show the full angular dependence of the blr @xmath86 intensity at different distances @xmath87 along the jet for the case where the blr is a thin shell at @xmath88 pc . for distribution ( ii ) we shall calculate the intensity by integrating the emissivity along lines of sight through the blr sketched in fig .
a fraction @xmath89 of the central luminosity would then be re - processed in @xmath90 . since this spherical shell has volume @xmath91 the emissivity ( in erg s@xmath56 sr@xmath56 @xmath92 ) of re - processed radiation inside the blr at radius @xmath93 is @xmath94 for @xmath95 and @xmath96 , where @xmath97 and @xmath98 is the number density and the radius of clouds at @xmath99 , respectively , the optical depth of the entire blr to central uv radiation is @xmath100 from which we obtain the constant @xmath101 for any assumed value of @xmath76 .
note that we shall use the power - law exponents preferred by kaspi and netzer @xcite , @xmath102 and @xmath103 .
we treat separately the three possible emission region locations shown in fig .
[ fig4 ] : ( a ) blob 1 inside the cavity inside the blr , ( b ) blob 2 located within the blr region , and ( c ) blob 3 is outside the blr .
the intensity of radiation at angle @xmath2 to the jet axis at a position @xmath87 _ within _ the blr region ( e.g. blob 2 in fig .
[ fig4 ] ) is given by @xmath104 where @xmath105 @xmath106 and @xmath107 the total ( angle - integrated ) energy density of blr is : @xmath108 and is plotted in fig .
[ fig6 ] as a function of distance along the jet . the energy density of blr photons clearly depends also on the distribution of blr clouds . at @xmath109 ( distribution i ) and @xmath110 ( distribution ii )
the blr photon energy density approaches the limit @xmath111 ( dashed line in fig .
[ fig6 ] ) . for distribution ( i ) , where we have a geometrically thin spherical shell for the blr , the energy density peaks sharply at @xmath65 as found also by ghisellini and madau @xcite . in the case of distribution ( ii )
the energy density peaks near @xmath99 , but the curves are smoother as a result of the finite thickness of the blr .
both cases are idealizations and there is probably a spread in cloud size at any given radius . from fig .
[ fig6 ] we see that it is very important to know the right distribution of blr clouds since there can be a difference of more than two orders of magnitude between the energy density of photons calculated for the same blr luminosity but different blr cloud distributions . in fig .
[ fig8](a ) we show the optical depth from various emission region heights , @xmath112 , to infinity for absorption of 100 gev100 tev @xmath0-rays in the radiation field of the blr for distribution ( i ) where the blr clouds are located in a geometrically thin shell at radius 0.01 pc .
the optical depth is plotted for a blr luminosity of @xmath113 erg / s , but can be linearly scaled for other blr luminosities .
the anisotropy of the blr radiation field is reflected in the shapes of the curves as one goes to higher @xmath0-ray energies .
[ fig8](b ) gives the optical depth for cloud distribution ( ii ) , with the inner radius at 0.01 pc ( the same as @xmath65 in part a ) and outer radius 0.4 pc . in this case , while at @xmath49 pc the optical depth is less than for distribution ( i ) , @xmath114 drops off more slowly with @xmath112 , such that at only slightly larger distances the optical depth becomes orders of magnitude higher than for distribution ( i ) until @xmath115 is reached .
this is in part because the possibility of near head - on collisions of @xmath0-rays with blr photons for @xmath116 for distribution ( ii ) large angles between directions of interacting photons means more collisions will be above the pair production threshold . fig .
[ fig18 ] shows the @xmath0-@xmath0 opacity for 10 tev photons from @xmath117 to infinity vs. @xmath64 for the two cases shown in fig .
[ fig8 ] . for
blazars with a very low blr luminosity , 10 tev photons can probably escape through the blr .
however , in quasars blr typically have @xmath118 erg / s , and so absorption of 10 tev @xmath0-rays can be quite important if the emission region is inside the blr ( distribution i ) or is close to @xmath67 ( distribution ii ) .
we comment here on a very intriguing problem , namely the absence of a ly absorption edge in quasars @xcite .
it was expected that blr clouds situated along the line of sight should produce a sharp ly absorption edge in more than 30% of quasars @xcite . however ,
maiolino et al .
@xcite have noted that none of the 101 quasars in the sample analyzed by zheng et al .
@xcite show this feature .
there are two possible explanations for the missing edge : ( a ) blr clouds have a sufficiently large component of their random velocity along our line of sight to result in a broadening of the ly edge and thereby making it difficult to observe , or ( b ) the blr does not have a spherically symmetric distribution around the ionizing accretion disk , such that there are no blr clouds intersecting our line of sight to the disk , and maiolino et al .
@xcite suggest a flattened geometry could do this .
such a flattening in the cloud distribution could produce weaker densities of blr photons and so would impede less the escape of tev photons from the central region .
however , we adopt a spherical distribution of clouds , and instead interpret the absence of an observed ly absorption edge in 101 quasars to suggest that the blr clouds must have a covering factor @xmath119 or less ( 95% upper limit ; for a covering factor 0.01 the probability of a blr cloud being in the line of sight to the disk in none of 101 observations is 36% , quite consistent with the observations ) . for
a given covering factor , @xmath120 , the fraction of the disk luminosity intercepted by the blr must then be less than @xmath121 , and justifies our use of @xmath81 .
nevertheless , @xmath122 have also been widely used , and would lead to photon - photon pair production opacities a factor of 10 higher than those presented here .
chiaberge et al .
@xcite have suggested that tori do not exist in fr - i sources , or they must have a very thin geometry .
this conclusion results from the correlation found between optical and radio cores of some fr - i sources .
however , the sample used in this analysis represents only just under half of the fr - i galaxies tabled by zirbel and baum @xcite , and one needs additional observations of the polarized optical spectrum to reach a firm conclusion about this @xcite .
nevertheless , these results could be interpreted as suggesting that perhaps there is a correlation between a flat infrared torus and a flat distribution of blr clouds as suggested by maiolino @xcite .
we do not know what mechanism , other than accretion , could cause such a flattening of the blr and the torus .
however , such a flattening might be understood in terms of a non - homogeneous spherical shell of blr or dust in which the polar caps above and below the disk plane are much more diluted than the equatorial belt .
zier and biermann @xcite have shown that the caps of a spherical layer of dust could be diluted so much , that even a dense torus would have a doughnut shape .
we have already discussed some aspects related to the infrared radiation from ubiquitous dusty tori in quasars .
unification schemes for agn generally involve the following components : kerr black hole , a relativistic accretion disk , jets , broad line clouds and a dusty torus . for blazars
the situation appears at first sight to be a little different , and we wonder why that might be .
there are two issues which should be considered in connection with dusty tori in blazars .
firstly , since there is no direct evidence of any thermal emission from tori we should consider what indirect evidence ( at any wavelength ) there is about the existence of dust in the centres of host galaxies .
secondly , we should consider what kind of torus geometry would fit observations made at different wavelengths best .
since unification schemes propose that fr - i and bl lacs are related , fr - i sources being the mis - oriented counterparts of bl lacs @xcite , we have searched the literature for information about dust structures associated with blazars / bl lacs / fr - i .
we note that blazejowski et al .
@xcite required an external ir photon field from a torus scattered by relativistic jet electrons , in addition to those in the ssc model , in order to fit @xmath0-ray spectra of ovv quasars observed by egret .
a key to the existence of tori in blazars could come from establishing a link between the torus and the blr .
for example , when there is no detection of direct optical emission , the interpretation of spectro - polarimetric data on fr - i objects with strong evidence for infrared obscuration ( @xcite and references therein ) could suggest the existence of a blr hidden by a thick torus . in this case , the observed polarized broad lines would be the result of scattering into the line of sight by free electrons in zones whose shape and orientation is determined by the torus inner geometry .
hence , the detection of free electron scattering regions should be a good diagnostic of torus geometry .
falcke et al .
@xcite proposed that the opening angle of the torus might play a critical role in the fr - i and fr - ii dichotomy a closed torus covering a large fraction of @xmath123 steradians , as seen from the black hole , would obscure the internal activity of fr - i objects .
de koff et al .
@xcite have discussed how the properties of the dust depend upon the radio properties of the object , and found that fr - ii , which are powerful radio galaxies , have a rather clumpy dust distribution , and they suggest that this translates into a large opening angle of the torus .
a less powerful radio jet ( such as those in fr - i ) would only slowly disperse the dust , and this type of obscuration might contribute to the deceleration of fr - i jets by entraining the material from the torus .
we consider first the fr - i galaxy centaurus a. alexander et al .
@xcite explain the ir spectrum of cen a using a combined model of infrared emission from startbursts , cirrus clouds and an agn - type torus .
a compact optically thick torus with @xmath124 pc and an opening angle of @xmath125 would account for the observed flux .
centaurus a appears as a concentrated ir source in the h band @xcite within a scale of tens of pc .
no optical emission has been observed from the central source and , as falcke et al .
@xcite suggested , this source may have a closed torus hiding the blr which has not been detected even in polarized flux .
israel et al .
@xcite , rydbeck et al .
@xcite and turner et al .
@xcite concluded that there is a dense nuclear torus with diameter @xmath126 pc .
it is also interesting that bryant et al .
@xcite show a clear picture of a warped dust lane obscuring the nuclear source the radio jet being perpendicular to a large - scale torus which seems to be inclined with respect to the plane of the dust - lane .
turning now to other fr - i radio galaxies , 3c218 ( hydra a ) also shows evidence for a nuclear obscuration @xcite .
3c270 is another fr - i radio galaxy for which a broad @xmath60 emission line has been detected @xcite .
ferrarese et al .
@xcite proposed that its nucleus is surrounded by a dust torus with a diameter of 120 pc and optical depth @xmath127 .
an inclination angle of @xmath128 between the axis of the torus and the line of sight would be in accordance with the detection of un - obscured nuclear optical emission @xcite .
possible evidence against tori in blazars comes from the fr - i radio galaxy m87 , which seems to have a very low ir flux which can not be explained with a standard torus model @xcite .
the torus in this source could be very diluted , and heated less as a result of the extremely low accretion activity of m87 @xcite such that its emission is dwarfed by the jet emission . at kiloparsec scales , starburst activity dominates the infrared emission , but as one goes deeper into the nucleus of the galaxy , de koff et al .
@xcite found that well organized dust structures were present in fr - i galaxies in the 3cr catalog , and this could indicate that the flow of matter towards the nucleus is rather steady in fr - i , allowing the formation of distinct torus features .
these structures tended to be sharply defined small - scales disks with radii less than @xmath129 kpc .
the detection of large - scale dusty features ( sometimes shaped as bars or dust - lanes ) suggests that there could be an association between a small - scale infrared torus , such as those in seyfert galaxies , and a large - scale torus sometimes identified with the dust structure of the galaxy .
the structure of the nuclear torus could then depend on the large - scale dust distribution
well - organized kpc dust structures could extend inward towards smaller scales ( small - scale torus surrounded by large - scale torus ) .
we define this as a symbiosis between the large and small scale dusty features @xcite .
the above discussion leads us to the conclusion that dust could be present everywhere , even in fr - i objects where the torus can not easily be detected .
since fr - i and bl lacs are thought to be similar ( apart from orientation ) and there is some evidence of blr and torus activity in fr - i , we postulate that the linkage between blr and tori in quasars discussed by chiaberge et al .
@xcite may apply also in blazars .
we shall calculate next the optical depth for @xmath0-@xmath0 absorption in the infrared radiation field of tori .
we know that the ir emission from the torus is strongly related to the activity of the central object .
the inner radius depends on @xmath130 , and is given by the sublimation radius of the dust @xcite : @xmath131 where @xmath132 erg / s , and @xmath133 is the dust temperature in units of 1500 k ( the dust sublimation temperature is taken to be 1500 k ) .
the inner torus radius is plotted in fig .
[ fig11 ] versus disk luminosity for three dust temperatures .
we start with a torus centred on the black hole , symmetric about the jet axis , and having a rectangular cross section with full height @xmath134 , and inner and outer radii @xmath135 and @xmath136 .
we shall discuss how @xmath114 changes for different scales of tori , e.g. , simulating an open torus by having either a rectangular cross section or a rectangular cross section with the inner edges cut away at angle @xmath137 ( see fig .
[ fig3 ] ) .
as discussed earlier , in the context of the symbiosis between jets and accretion disks , an increase in jet power could be at the expense of disk luminosity .
a lower disk luminosity at uv frequencies would reduce the heating of the inner surface of the dusty torus , causing the inner radius of the torus , given by eq .
[ 14 ] , to be small ( fig .
[ fig11 ] ) .
this implies that fat dusty tori ( @xmath138 ) could exist in agn with very low central activity such as blazars , and this would be in accordance with the closed torus model discussed by falcke et al .
@xcite . on the other hand , for a `` cold torus '' with @xmath139
k , the inner radius would be far from the nucleus for a given luminosity ( see fig . [ fig11 ] ) .
such inefficient heating of the dust would probably describe best the torus in m87 which seems to be a peculiar object in that it has an extremely low mass accretion rate .
due to its size , the energy density of the infrared photons remains fairly uniform inside the torus .
a flat torus ( i.e. , @xmath140 ) could approximate the models proposed by chiaberge et al .
@xcite ; see the discussion from the previous section and comments by maiolino et al .
@xcite about a flattened blr distributions .
although flat tori have not been detected , and we are uncertain about the stability of such a configuration of dust , we believe that it is an interesting possibility . for simplicity we model in this section a torus with a single black body temperature such that the ir intensity is constant within the solid angle subtended by the torus .
we shall extend the work of protheroe & biermann @xcite for different torus geometries .
[ fig3 ] illustrates the geometry for interaction , at points a and b , of @xmath0-rays with ir photons emitted from the surfaces of the torus .
accretion rates are smaller in blazars than in quasars . since the thickness of the torus is related to the accretion inflow and disk luminosity ( through radiation pressure effects ) @xcite one expects tori in fr - i to be thinner , diluted , and/or less efficiently heated as is probably the case in m87 .
therefore , we have varied @xmath134 between @xmath141 pc and @xmath142 pc , @xmath143 between @xmath144 pc and 1 pc , and @xmath145 between 2 pc and 10 pc . the optical depth from @xmath146 to infinity , for @xmath0-rays traveling along the jet axis ,
is shown in fig .
[ fig9 ] where we have compared the case of a torus with @xmath147 pc ( solid curves ) with a torus having @xmath148 pc ( dashed curves ) for three different torus heights @xmath134 .
we find that more @xmath0-rays with energies around @xmath149 gev would be absorbed by @xmath0-@xmath0 interactions when the torus comes closer to the central source .
we also find that an open torus geometry ( @xmath150 ) can significantly modify @xmath114 as shown in fig .
[ fig9vary](a ) where it is again seen to have the greatest effect near the pair production threshold .
if the torus has a large outer radius , then more ir photons from the upper surface of the torus are available for interaction with @xmath0-rays along the jet .
for example , for an extended torus with @xmath151 pc the absorption of photons with energy above 2 tev is larger ( compare dotted line with upper solid curve in fig .
[ fig9vary]a ) . as noted previously by protheroe and biermann @xcite
, we can see that for photon energies above about several hundreds of gev the opacity is large enough such that no tev @xmath0-rays can emerge if the source is near the centre of the torus .
risaliti et al .
@xcite have shown that a torus is stable if the mass of the dust in the torus does not exceed the dynamical mass flowing toward the center .
this constrains the outer torus radius , @xmath145 , to have values less than 10 pc for a column density of dust along the line of sight that intersects the obscuring medium of @xmath152 @xmath153 .
fr - i objects could have diluted tori with column densities of @xmath154 @xmath153 , or lower , but still be optically thick , and this would impose a upper limit of @xmath155 pc for the torus outer radius . however , maiolino et al .
@xcite interpret their data as suggesting that the inner parts of tori are much denser than the outer parts , and find a gradient in the torus covering factor that would give @xmath156 pc for seyfert galaxies .
the covering factor plays a crucial role in these models , as well as the gas mass enclosed within the torus @xcite . for simplicity , we take a value of 10 pc as the maximum radius of the dust torus which could be relevant to our analysis . in fig . [ fig9vary](b ) we show how the optical depth depends on the distance of @xmath0-ray emission region along the jet .
so far , we have assumed that the torus radiates as a perfect black body .
however , if the torus is patchy or not optically thick at ir wavelengths we may have diluted black body radiation , @xmath157 erg/@xmath158 , where @xmath159 is the radiation constant .
all our curves would be multiplied by @xmath160 in this case .
alternatively , for a poor heating mechanism related to a low luminosity of the accretion disk the dust could be heated to lower temperatures , e.g. @xmath161 k , and this results in reducing the gev to tev opacity considerably , with the pair production threshold @xmath0-ray energy increasing as @xmath162 as illustrated in fig .
[ fig9vary](b ) by the dotted curves .
we note however , that blazejowski et al .
@xcite require @xmath163 to fit @xmath0-ray spectra of ovv quasars with an external inverse compton model , and this still allows for significant tev @xmath0-ray absorption in this type of blazar .
extremely low values of @xmath164 may apply in objects such as m87 .
the adj model gives a simplified approach to the symbiosis between disks and jets .
it could explain the low central activity in some blazars , and for very low mass accretion rates an adj could transform into an edaf or an adaf .
the strong variability observed in blazars from radio to @xmath0-ray frequencies could be a consequence of the feeding mechanism of the jet at the boundary layer .
a local perturbation produced in the boundary layer , such as a rapid variation of @xmath11 or @xmath165 , would propagate through the nozzle and along the jet causing transient phenomena .
as the jet extracts more energy from the disk , this could leave less energy to be dissipated in the disk as radiation . the jet / disk symbiosis
then mainly modifies the energetics in the central parsec of agn : increasing the power of the jet could weaken the disk which therefore would not efficiently ionize the blr clouds or efficiently heat the dust in the torus , and so would affect important target photon fields for interactions of @xmath0-rays ( and also accelerated electrons and protons ) along the jet .
[ fig7 ] summarizes our results in a way that makes it easy to see the relative importance of the three radiation fields to the absorption of high energy @xmath0-rays .
we show the optical depth as a function of height of the emission region above the plane of the disk for three @xmath0-ray energies spanning the range currently explored with optical cherenkov telescopes : ( a ) 0.5 tev which is close to the threshold of present telescopes , ( b ) 2.5 tev , ( c ) 12.5 tev typically observed during flaring activity in nearby blazars .
the shaded areas give the range in optical depth for the models discussed earlier , but normalized for an accretion disk luminosity @xmath166 erg s@xmath56 , a broad line region luminosity @xmath167 erg s@xmath56 , and torus ir dilution factor @xmath168 .
we see that at all three energies , all three radiation fields can play a very important role .
we shall consider each radiation field in turn , first for quasars and then for blazars . in quasars ,
with typically @xmath169 erg s@xmath56 the accretion disk radiation will surely cut off @xmath0-rays above 0.5 tev if the emission region is below @xmath1700.03 pc ( depending on @xmath171 ) . at 2.5
tev and 12.5 tev the corresponding minimum emission region heights to avoid absorption in quasars are @xmath1720.08 pc and @xmath1730.2 pc .
the blr is active in quasars , and is likely to have @xmath118 erg s@xmath56 such that the emission region would have to be above 0.030.3 pc ( depending on blr cloud distribution ) at all three energies . for a luminous disk as in quasars
, one would expect an active torus to cut off tev energy @xmath0-rays unless the emission region were well above the torus : @xmath1742 pc ( 0.5 tev , note @xmath175 can be small at the energy ) , @xmath17610 pc ( 2.5 tev ) , and @xmath17730 pc ( 12.5 tev ) , depending on torus geometry .
all known quasars are at a distance such that multi - tev @xmath0-rays are expected to be absorbed in the infrared background radiation during their propagation to earth .
however , observations at energies below 1 tev are possible with current atmospheric cherenkov telescopes , and multi - gev observations will be made with glast @xcite within a few years .
future measurements of the shape of the @xmath0-ray spectrum of quasars at the cut - off might have something to say about which radiation field is responsible for the cut - off , and hence the location of the emission region , although this is likely to be very difficult ( compare the optical depth vs. energy in the disk radiation , blr radiation and torus radiation in figs . 2 , 6 and 11 ) .
the situation is somewhat different in blazars . here
, we can have significantly less powerful accretion disks than in quasars .
for example , to avoid attenuation of @xmath0-rays in the disk radiation for a disk luminosity of @xmath113@xmath178 erg s@xmath56 one would need to be above @xmath1790.005 pc ( 0.5 tev ) , and above @xmath1800.03 pc ( @xmath181 tev ) , depending on @xmath171 . note that at all energies the optical depth in the blr field becomes less than unity if the blr luminosity is lower than @xmath182 erg s@xmath56 .
however , for @xmath183@xmath113 erg s@xmath56 the @xmath0-ray emission region would need to be above 0.030.3 pc ( depending on blr cloud distribution ) , as for quasars , to avoid attenuation . in the case of the torus radiation ,
if blazars have active tori , they could present a serious problem for the escape of tev @xmath0-rays from emission regions which are not well above the torus ( as in the case of quasars ) .
because of the low disk luminosity , the torus inner radius would probably be small ( @xmath184 pc for @xmath185 erg s@xmath56 , see eq . 11 ) , and the torus would be likely to be small , generally , and the optical depth would probably be at or below the lower boundary of that shown .
inefficient heating could also result in dilution of the torus radiation ( @xmath186 ) or in a lower temperature , and consequently lower optical depths ( see e.g. fig .
we have demonstrated that the external radiation fields present within the parsec scale of agn may create problems for the escape of high energy @xmath0-rays .
the effect is dramatic for quasars where the densities of external radiation fields are much higher than in blazars .
the symbiosis between the accretion disk and jets could result in lower radiation fields around the base of the jet during flaring activity if the jet takes a higher fraction of the matter entering the disk , and this could minimize the dramatic effect of the @xmath0-ray absorption .
however , although heated by the disk , the torus ir field would not react rapidly to changes in disk luminosity because of its larger scale , and so the escape of tev @xmath0-rays would require the emission region to be located at parsec distances from the black hole , far above the torus unless the torus is very inefficiently heated although even in this case one can expect some attenuation .
this is in accordance with conclusion of protheroe and biermann @xcite , and also the recent work of jorstad et al .
@xcite who concluded that the temporal connection between the radio and @xmath0-ray variations suggests that the @xmath0-rays appear to originate far away from the radio core . in conclusion , the origin of the 100 gev-1000 gev @xmath0-ray photons must be above the inner part of blr region for any quasar .
blazars usually have low central luminosities , and this translates into a having a low blr luminosity , and possibly a closed torus .
mrk 421 seems to have a blr with an extremely low luminosity @xmath187 erg / s @xcite and so could be transparent to tev @xmath0-rays .
however , tev @xmath0-rays would need to be produced far above the torus unless the torus is very inefficiently heated .
farther along the jet external radiation fields become less important . in this case
the radiation produced inside the emission region in the jet ( e.g. synchrotron emission ) becomes the dominant radiation field for photon - photon pair production by @xmath0-rays produced in the jet .
we thank peter biermann and geoff bicknell for useful discussions .
this work was supported by an australian research council grant to rjp . | the radiation fields external to the jets and originating from within a few parsecs from the black hole , are discussed in this paper .
they are the direct radiation from an accretion disk in symbiosis with jets , the radiation field from the broad line region ( blr ) surrounding the accretion disk , and the infrared radiation from a dusty torus .
the jet / disk symbiosis modifies the energetics in the central parsec of agn such that for a given accretion rate , a powerful jet would occur at the expense of the disk luminosity , and consequently the disk would less efficiently ionize the brl clouds or heat the dust in the torus , thereby affecting potentially important target photon fields for interactions of gamma - rays , accelerated electrons and protons along the jet .
motivated by unification schemes of active galactic nuclei , we briefly review the evidence for the existence of broad line regions and small - scale dust tori in bl lacs and fanaroff - riley class i ( fr - i ) radio galaxies .
we propose that an existing jet - accretion disk symbiosis can be extrapolated to provide a large scale - symbiosis between other important dusty constituents of the blazar / fr - i family . in the present paper , we discuss in the context of this symbiosis interactions of gev and tev gamma - rays produced in the jet with the various radiation fields external to the jet in quasars and blazars , taking account the anisotropy of the radiation . , submitted + * radiation fields of disk , blr and torus in quasars and blazars : + implications for gamma - ray absorption * + alina - c .
donea and r.j .
protheroe + department of physics and mathematical physics + the university of adelaide , adelaide , sa 5005 , australia * pacs codes : * 98.54.aj quasars 98.54.cm active and peculiar galaxies 98.70.rz gamma - ray sources * keywords : * accretion disk , broad line region , torus , gamma - ray , agn , photon - photon pair production |
over the past few years the idea of integrating the concepts of spintronics and molecular electronics has developed into a new research field dubbed _
molecular spintronics_.@xcite progress has not only been stimulated by technological interests but has also been accompanied by the realization that magnetic single - molecule transistors exhibit various fundamental quantum phenomena.@xcite among many promising ideas discussed in the literature , particular attention has been paid to current - induced spin reading and writing , spin relaxation , entanglement , quantum computation , and kondo correlations . an experimental realization of spintronics devices may be achieved by using single - molecule magnets in combination with metallic ( nonmagnetic or ferromagnetic ) leads . for molecular - memory applications ,
long spin - relaxation times are advantageous , which may be realized in molecules with large magnetic anisotropy , such as molecules based on @xmath0 , @xmath1 , and @xmath2.@xcite controlling and detecting the molecular spin by means of electronic tunneling into source and drain electrodes poses a major challenge .
while some approaches rely on break junctions , others are based on a scanning tunneling microscope . in both cases ,
the coupling between the molecule and the leads can vary by several orders of magnitude , thus giving rise to strikingly different transport regimes . in the regime of weak molecule - lead coupling , many experimental features such as coulomb blockade
, spin blockade , sequential tunneling , and cotunneling can be described within a master - equation or rate - equation approach treating the electronic tunneling perturbatively.@xcite however , for strong coupling , low - order perturbation theory breaks down .
it is then advantageous to treat the electronic tunneling exactly , at the price of introducing approximations elsewhere .
recently , the kondo effect in single - molecule magnets with easy - axis anisotropy has been studied by romeike _ _ et al.__@xcite their model describes an anisotropic spin coupled to metallic electrodes by an exchange interaction , in the absence of a bias voltage .
this differs from the model studied here , in which the electronic spin in the relevant molecular orbital is coupled to an additional local , anisotropic spin via an exchange interaction @xmath3 , i.e. , charge fluctuations are explicitly taken into account .
in addition , we include a nonzero bias voltage .
the presence of a kondo effect for an anisotropic spin is at first glance surprising since it requires two approximately degenerate low - energy spin states connected by a term in the hamiltonian .
the simplest hamiltonian for an anisotropic spin @xmath4 with @xmath5 exchange - coupled to an electronic spin @xmath6 @xmath7 does not provide such a term . using a renormalization - group approach , romeike _ _
et al.__@xcite could show that quantum tunneling of the magnetic moment , which is described by higher - order anisotropy terms not included in eq .
( [ basich ] ) , may give rise to a kondo peak in the linear conductance , centered at zero bias voltage .
the kondo temperature is found to depend strongly on the ratio of the applied magnetic field and the anisotropy barrier .
further , gonzlez _ _ et al.__@xcite have derived a kondo hamiltonian of the type studied in ref . from an electronic model
they have shown that a transverse magnetic field can induce or quench the kondo effect .
this is due to berry - phase interference between different quantum tunneling paths of the spin .
koerting _ _ et al.__@xcite consider the nonequilibrium kondo effect for a double quantum dot with four leads . by removing the leads coupled to one of the dots one would obtain a model similar to ours .
the main difference is that we include charge fluctuations on the dot which is coupled to the leads , whereas koerting _ _ et al.__@xcite work in the regime of weak tunneling and coulomb blockade , where both quantum dots act as local spins . in the present paper
, we address the question of spin - dependent resonant tunneling and kondo tunneling through molecular magnets .
as noted above , we consider a resonant - tunneling model where the electron spin on the quantum dot or molecule is coupled to an additional local , anisotropic spin via an exchange interaction @xmath3 .
we assume this interaction to be weak , which allows us to employ perturbation theory for small @xmath3 .
the stationary current through the left and right leads is identical and is related to the local electronic spectral function @xmath8 on the molecule by the meir - wingreen formula@xcite @xmath9 \ , a_{\sigma}(\omega),\ ] ] where @xmath10 is the spin , @xmath11 is the broadening of the molecular level due to the hybridization with lead @xmath12 , to be defined below , and @xmath13 denotes the fermi distribution function of lead @xmath14 .
the spectral function is determined by the imaginary part of the retarded green s function , @xmath15 .
if transport is dominated by a single molecular level of energy @xmath16 , the coupling to the leads gives rise to a lorentzian form of the spectral function , @xmath17 $ ] with @xmath18 , which manifests itself as a peak in the differential conductance .
for single - molecule devices , the excitation of additional degrees of freedom due to the electronic tunneling is expected to translate into additional characteristic features in the current .
we consider two complementary situations .
the first is the case of arbitrary gate and bias voltages but excluding the region where the kondo contribution to the current is large . within a master - equation approach treating the local exchange interaction
perturbatively to second order , we calculate the transition rates between local - spin states , showing that the spin can be driven out of equilibrium even for strong molecule - lead hybridization .
signatures of inelastic tunneling such as the fine - structure splitting of the differential - conductance peaks persist in the regime where the hybridization energy exceeds the thermal energy .
the second case concerns the regime of a large kondo contribution to the differential conductance , which only occurs for small bias voltages on the order of @xmath19 , as we shall see . here
, transport is studied using a diagrammatic approach .
we consider the case that the molecular orbital is far from resonance so that the resonant - tunneling contributions are negligible .
in addition , this allows us to obtain analytical expressions .
we find that the magnetic anisotropy gives rise to the appearance of two kondo peaks in the differential conductance at finite bias voltages @xmath20 .
this intrinsically nonequilibrium kondo effect is quite different from the zero - bias peak studied by romeike _
et al._,@xcite which relies on higher - order anisotropies absent from our model . in our case ,
the magnetic anisotropy acts like a magnetic field in that it gives rise to a splitting of the kondo peak .
furthermore , we find a suppression of the differential conductance with @xmath21 .
the paper is organized as follows . in sec .
[ sectionmodel ] , we introduce our model .
section [ sectionmasterequation ] considers transport within a master - equation approach , which allows us to study magnetic nonequilibrium phenomena , whereas sec .
[ sectionkondo ] considers a diagrammatic approach , which applies to the kondo regime . in sec .
[ sectionconclusion ] we summarize and discuss our results further .
some detailed calculations are relegated to appendices .
we consider a magnetic molecule coupled to two metallic leads .
electronic tunneling through the junction is assumed to involve a single orbital with energy @xmath16 and spin @xmath6 that is coupled to a local spin @xmath22 via exchange interaction .
the model is described by the hamiltonian @xmath23 where @xmath24 is the resonant - tunneling hamiltonian , @xmath25 with @xmath26 is the exchange interaction between the electrons in the molecular orbital and the local spin @xmath4 , and @xmath27 describes the easy - axis magnetic anisotropy of the local spin .
we choose the _ z _ axis as the easy axis .
here , @xmath28 creates an electron with spin @xmath10 and energy @xmath16 on the molecule , while @xmath29 creates an electron with energy @xmath30 , wave vector @xmath31 , and spin @xmath10 in lead @xmath14 , which is considered a noninteracting electron gas .
the vector @xmath32 denotes the pauli matrices . in break junctions produced by electromigration
, the onsite energy @xmath16 can be tuned by applying a gate voltage.@xcite
the presence of strong coupling between the molecule and the leads prevents us from treating the hybridization term in eq .
( [ resonant - tunneling ] ) perturbatively .
however , since the hamiltonian becomes bilinear in the limit of vanishing exchange coupling , @xmath33 , our strategy is to diagonalize @xmath34 exactly while treating @xmath35 as a perturbation up to second order .
this approach allows us to study the nonequilibrium dynamics of the molecular spin at finite bias voltages for arbitrary molecule - lead coupling strengths , provided that kondo correlations do not lead to a diverging contribution from higher - order terms in the expansion .
we start by rewriting @xmath36 in terms of new operators,@xcite @xmath37 where @xmath38 and @xmath39 for simplicity we assume real tunneling amplitudes @xmath40 . in terms of the new operators , the exchange interaction assumes the form @xmath41 the time evolution of the density matrix @xmath42 of the full system is described by the von neumann equation , @xmath43 $ ] .
the degrees of freedom of the local spin are described by the reduced density matrix @xmath44 which is obtained by tracing out all electronic degrees of freedom . assuming that the large electronic subsystem , which acts as a spin reservoir , is weakly perturbed by the exchange coupling
, we replace the full density matrix by the direct product @xmath45 .
we need a further approximation for the electronic density matrix @xmath46 .
we assume that different chemical potentials @xmath47 ( @xmath48 ) are imposed for the left ( right ) lead far from the junction .
it would thus be natural to assume fermi distributions @xmath49 for the physical electrons created by @xmath50 .
however , we need to make a reasonable assumption on the transformed
fermions appearing in eqs.([h0c.1 ] ) and ( [ h_j ] ) .
since @xmath51 ( @xmath52 ) creates an electron in a state with vanishing probability density far from the junction in the right ( left ) lead , we assume the occupation numbers of these states to be described by @xmath53 . making use of the markov approximation that @xmath54 changes slowly on the time scale of electronic relaxation , we obtain @xmath55 \right].\ ] ] here , operators @xmath56 with an explicit time argument , including @xmath57 , are in the interaction picture , @xmath58 .
note that second - order perturbation theory in the exchange coupling gives the first non - vanishing correction to the conductance , since the expectation value @xmath59 and thus all first - order terms vanish exactly due to symmetry .
we are interested in the stationary state .
the stationary density matrix @xmath54 has to be diagonal in the basis of eigenstates @xmath60 of @xmath61 , since the full hamiltonian @xmath62 is invariant under rotation about the _
z_-axis in spin space .
inserting eq .
( [ h_j ] ) into eq .
( [ iterated - rho ] ) we thus obtain a pauli master equation , also called rate equations , of the form @xmath63 for the occupation probabilities @xmath64 of spin states @xmath60 in the stationary state .
the transition rates read @xmath65k_2)|^2 \nonumber \\ & & { } \times \left[1-f_{\alpha}(\omega ) \right ] f_{\alpha'}(\omega-[\pm 2m+1]k_2).\end{aligned}\ ] ] the spectral functions are given by @xmath66 with @xmath67 and @xmath68 .
the densities of states for the leads , @xmath69 , are taken as constants .
compared to eq .
( [ transformation3 ] ) we have approximated the self - energy part of @xmath70 by a constant and absorbed a factor @xmath71 . at zero temperature , the integrals can be evaluated analytically . in the limit of large bias voltages ,
the rates approach the constant value @xmath72 } \nonumber \\ & & { } \times |\langle m \pm 1 | s^\pm | m \rangle|^2.\end{aligned}\ ] ] on the other hand , at zero bias only the rates involving the absorption of energy are finite , whereas the emission rates vanish . solving eq .
( [ rate - eq ] ) allows us to compute the differential conductance of the molecular junction .
the current operator of lead @xmath14 reads @xmath73 in order to compute the spin - dependent contribution to the expectation value @xmath74 of the total current , we use the iterative solution of the von neumann equation , @xmath75 \right].\ ] ] a term containing @xmath76 has dropped out here since it is linear in @xmath35 and thus vanishes upon taking the trace . making use of the markov approximation we find @xmath77 , h_j(t '' )
\right ] \rho(t)\end{aligned}\ ] ] for the second - order term . carrying out the time integrals and evaluating the spin sums as explained in appendix [ appcurrent ]
, we obtain @xmath78 \big ) } \nonumber \\ & & { } \times \sum_m p_m\ , \bigg\ { \frac{1}{2}\ , |\langle m-1 | s^- | m \rangle|^2 i_{\alpha\alpha'\alpha''}([-2m+1]k_2 ) \nonumber \\ & & \quad { } + \frac{1}{2}\ , |\langle m+1 | s^+ | m \rangle|^2 i_{\alpha\alpha'\alpha''}([2m+1]k_2 ) \nonumber \\ & & \quad { } + |\langle m | s^z | m \rangle|^2 i_{\alpha\alpha'\alpha''}(0)\bigg\}\end{aligned}\ ] ] with @xmath79 f_{\alpha'}(\omega - e ) \nonumber \\ & & \quad { } - |\tilde{\nu}_{\alpha}(\omega)|^2 \ , |\tilde{\nu}_{\alpha'}(\omega+e)|^2 \ , f_{\alpha}(\omega ) \left[1-f_{\alpha'}(\omega+e)\right ] \big\}. \nonumber \\ & & { } \end{aligned}\ ] ] equations ( [ currentj ] ) and ( [ currentiaaa ] ) give the first non - vanishing correction to the zero - order current @xmath80 , which is obtained from the meir - wingreen formula [ eq .
( [ eq1 ] ) ] by inserting the spectral function of the unperturbed system , @xmath81 $ ] .
note that eq .
( [ eq1 ] ) with @xmath82 is recovered by inserting the equilibrium density matrix @xmath83 and the current operator from eq .
( [ equationcurrentoperator ] ) into @xmath84 .
a simple interpretation of eq . ( [ currentj ] ) is possible for the special case of a local spin of length @xmath85 , for which the magnetic anisotropy @xmath86 is irrelevant and can be set to zero . for this case
we obtain @xmath87 ^ 3 \ , \left[f_{l}(\omega ) -f_{r}(\omega)\right].\end{aligned}\ ] ] here , the third power of the spectral function appears , since the current operator and the two exchange - interaction operators in eq .
( [ currentj0 ] ) are each bilinear in fermionic operators
. @xmath88{figa_symmetric.eps } \\
\textbf{(b ) } \\
\includegraphics[height=5.5cm , angle=0]{figb_symmetric.eps } \end{array}$ ] @xmath88{figa_asymmetric.eps } \\
\textbf{(b ) } \\
\includegraphics[height=5.5cm , angle=0]{figb_asymmetric.eps } \end{array}$ ] if the magnetic anisotropy is large compared to the hybridization energy , @xmath89 , the general expression for the current in eq . ( [ currentj ] )
simplifies to @xmath90 \big ) i_{\alpha\alpha'\alpha''}(0 ) \nonumber \\ & ~\times \sum_m p_m |\langle m | s^z | m \rangle|^2,\end{aligned}\ ] ] since the integrals @xmath91k_2)$ ] are negligible compared to @xmath92 . assuming symmetric capacitances , one finds , in the limit of large bias voltages , @xmath93 for the zero - order and second - order contribution , respectively . note that the ferromagnetic or antiferromagnetic sign of @xmath3 does not affect the results in the present approximation .
we first consider the situation of symmetric molecule - lead couplings and capacitances , i.e. , @xmath94 , @xmath95 , and @xmath96 , @xmath97 . figure [ fig_current_symmetric](a ) shows the current - voltage characteristics up to second order in @xmath3 for the case of a local spin of length @xmath98 .
the characteristic fine structure of the current step at the coulomb - blockade threshold is due to the second - order contribution , @xmath99 , whereas the main step is mostly coming from @xmath80 .
the fine structure persists as long as the hybridization energy @xmath100 remains small compared to the magnetic anisotropy @xmath86 .
note that the broadening of the steps is due to @xmath101 , and not to the temperature , for which we assume @xmath102 . for bias voltages below @xmath103 ,
the current and all magnetic excitations are thermally suppressed .
however , as soon as the chemical potential of one lead aligns with the resonance of the molecule , the current increases to its maximum value .
the current - induced magnetic transitions become energetically accessible at the same time , as shown in fig . [
fig_current_symmetric](b ) , resulting in nonequilibrium probabilities @xmath64 of the different spin states . in the limit of large bias
voltages all spin states are equally occupied , @xmath104 .
interestingly , the presence of magnetic anisotropy leads to negative differential conductance in the vicinity of @xmath105 .
the underlying mechanism shall be explained briefly . according to eq .
( [ refertothiseq ] ) , the magnetic states with maximum quantum numbers @xmath106 dominate the current since the current is proportional to the average @xmath107 .
each decrease in @xmath108 thus causes a decrease in the current .
therefore , the spin - dependent contribution to the current is large at low bias voltages , where @xmath109 and @xmath110 , and small at high bias voltages , where @xmath111 and @xmath112 .
@xmath113{levelscheme.eps } \\ \begin{array}{cc } \textbf{(b ) } & \hspace{0.2 cm } \textbf{(c ) } \\
\hspace{0.0cm}\includegraphics[height=4.2cm , angle=0]{current2d_symmetric.eps } & \hspace{0.0cm}\includegraphics[height=4.2cm , angle=0]{current2d_asymmetric.eps } \end{array } \end{array}$ ] we next consider the situation of strongly asymmetric molecule - lead couplings and capacitances , i.e. , @xmath114 , @xmath115 , and @xmath116 , @xmath117 . note that this regime is naturally realized in many experimental setups , whereas perfectly symmetric couplings are in general more difficult to achieve .
the current - voltage curves and the bias dependence of the magnetic excitation rates are shown in fig .
[ fig_current_asymmetric ] . due to the asymmetric coupling and @xmath118 , the current is suppressed for negative bias voltages .
however , the characteristic steps corresponding to excitations of the molecular spin reappear at positive bias
. only their abscissas are reduced by a factor of 2 , since the chemical potential of the left lead is now @xmath119 instead of @xmath120 .
the device thus acts as a rectifier .
note the small ohmic contribution with constant slope for large coupling @xmath100 in fig.[fig_current_asymmetric](a ) .
we return to this point below .
we finally turn to the full bias and gate - voltage dependence of the current . in both the symmetric and the asymmetric case , selection rules for the spin require changes in the magnetic quantum number by @xmath121 or @xmath122 , where @xmath121 corresponds to elastic and @xmath122 to inelastic scattering events , cf.fig .
[ fig_2dplots](a ) .
inelastic scattering processes appear as additional steps in the current and give rise to the magnetic fine structure in the two dimensional density plots of the second - order contribution to the current as a function of bias and gate voltages shown in figs .
[ fig_2dplots](b ) , ( c ) .
we can now understand the origin of the weak ohmic conduction seen in fig.[fig_current_asymmetric](a ) for @xmath123 .
what we are seeing is the tail of the current step at @xmath124 in fig .
[ fig_2dplots](c ) , which is considerably broadened for @xmath123 .
note that we here have @xmath125 , i.e. , we are only @xmath126 away from the step .
since this distance does not depend on the bias voltage , the conductivity is essentially constant , leading to ohmic behavior .
second - order perturbation theory in the exchange interaction @xmath3 fails , even for small @xmath3 , if the prefactors of higher - order terms diverge .
this is the case in the kondo regime .
logarithmic divergences of the conductance first appear in terms of _ third _ order in @xmath3.@xcite ( in this section , we assume antiferromagnetic exchange , @xmath127 . ) studying the emergence of kondo correlations thus requires to go beyond the second - order master - equation approach discussed in sec .
[ sectionmasterequation ] . for sufficiently small @xmath3 and sufficiently large thermal energies ,
the conductance is dominated by the third - order contribution , which we calculate in this section . at lower temperatures
, it would become necessary to resum the divergences to all orders in @xmath3.@xcite the total current through the molecule is related to the local electronic spectral function @xmath128 by the meir - wingreen formula , eq .
( [ eq1 ] ) , where @xmath129 denotes the fourier transform of the retarded single - particle green s function @xmath130 making use of the transformation defined in eqs .
( [ transformation1])([transformation3 ] ) requires to compute the finite - temperature time - ordered green s function @xmath131 our strategy is to expand @xmath132 in powers of @xmath3 . in order to obtain the current from the meir - wingreen formula , we need the imaginary part of the electronic green s function , @xmath133 where @xmath134 denotes the retarded green s function .
all non - vanishing diagrams of the local green s function up to third order in @xmath3 are shown in fig .
[ diagrams ] , following the notation of ref . .
we again consider the case of strongly asymmetric couplings , i.e. , latexmath:[$|t_l| \ll @xmath116 , @xmath136 .
as we shall see , eq .
( [ eq.img.2 ] ) is then dominated by the contribution from the right electrode , @xmath137 .
this allows us to describe the molecular degrees of freedom by a thermal equilibrium distribution function that is independent of the applied bias voltage and to obtain an analytical expression . , following ref . .
diagrams including fermion loops vanish exactly and are not shown .
spin averages are denoted by dotted lines.,height=151 ] the evaluation of eq .
( [ eq.img.2 ] ) is shown in appendix [ appgreensfunction ] .
we obtain @xmath138 with @xmath139 \bigg\},\end{aligned}\ ] ] where @xmath140 in the derivation we have assumed the molecular level to be far from resonance , i.e. , @xmath141 to be large compared to @xmath142 , @xmath143 , and @xmath100 , but still small compared to the band width @xmath144 of the leads .
details are discussed in appendix [ appgreensfunction ] .
we have also assumed @xmath145 , the significance of which will become clear in the following step . under these conditions , the resonant - tunneling contribution to the differential conductance , which we have studied in sec.[sectionmasterequation ] , is negligible compared to the kondo contribution . in the low - temperature limit @xmath102 ,
derivatives of the fermi functions with respect to the bias voltage become delta functions and the differential conductance simplifies to @xmath146 note that the argument @xmath147 of @xmath148 is @xmath149 .
the assumption @xmath145 made above thus corresponds to @xmath141 also being large compared to the bias , @xmath150 .
the spectral function has logarithmic divergences for @xmath151 at the transition energies of the molecule , @xmath152 , corresponding to virtual transitions between two magnetic states @xmath60 and @xmath153 .
one recovers the prefactor @xmath154 , see ref .
, for the case of spin @xmath85 and the ( then irrelevant ) anisotropy set to @xmath155 . in units of @xmath156 for different thermal energies @xmath143 ( in units of @xmath86 ) obtained from eq .
( [ eqforfigure ] ) as a function of bias voltage @xmath157 in units of @xmath158 . here
we assume a local molecular spin of length @xmath98 and choose @xmath159 , @xmath160 , and @xmath161 .
note that the parameters @xmath100 , @xmath16 , and @xmath3 leave the curves for @xmath162 ( in arbitrary units ) invariant except for changing the constant offset.,height=211 ] numerical results for nonzero temperatures are shown in fig .
[ fig_kondo ] .
the differential conductance @xmath162 diverges logarithmically for @xmath151 at critical bias voltages @xmath163 with @xmath164 since the emergence of kondo correlations requires the bias voltage to exceed the energy of the transition from the ground states , @xmath106 , to the first excited states , @xmath165 .
note that @xmath166 is symmetric for positive and negative bias , in spite of the highly asymmetric coupling since it is probing the electronic spectral function .
the situation is quite different from the case considered by romeike _
et al._,@xcite which concerns a zero - bias peak resulting from quantum tunneling between the two states @xmath167 and @xmath168 . in our case
, the splitting of the kondo peak as a consequence of magnetic anisotropy is more similar to the situation of a quantum dot in an external magnetic field with zeeman energy @xmath169 , where a zeeman splitting of the energy levels leads to the occurrence of two conductance peaks at @xmath170 in the kondo regime.@xcite at higher temperatures , @xmath171 , the two kondo peaks merge into a single peak centered at zero bias due to the thermal excitation of spin states with higher energy .
we now turn to the kondo temperature @xmath172 .
poor man s scaling for the equilibrium case results in @xmath173 because the matrix elements of @xmath174 between the two degenerate ground states of the local spin vanish for our model if @xmath175.@xcite since a kondo effect evidently does occur at nonzero bias , this result is clearly not sufficient .
a rough estimate of the kondo temperature @xmath172 can be obtained as the temperature for which the second - order and third - order terms become equal in eq .
( [ selfenergy2nd3rd ] ) .
we find @xmath176 $ ] , where @xmath14 is a number of the order of unity . in the limit @xmath177 , where the two peaks in fig .
[ fig_kondo ] would merge , we recover the result @xmath178 for an isotropic spin .
since we focus on the case of strongly asymmetric couplings , where the molecular degrees of freedom are in equilibrium with one of the two leads , the logarithmic divergences are cut off by temperature or the applied bias voltage , respectively , in our perturbative approach , see eq .
( [ selfenergy2nd3rd ] ) .
the divergence for @xmath151 is unphysical and would likely be removed by a resummation of higher - order terms .
by analogy to ref . , we conjecture that the divergence is ultimately cut off by a voltage - dependent spin - relaxation rate .
while we have so far discussed the dependence on the bias voltage , see fig .
[ fig_kondo ] , we now turn to the gate voltage .
the gate voltage shifts the on - site energy @xmath16 and thus enters the expression for the current through the square of the spectral function @xmath179 $ ] and the square of the factor @xmath180 $ ] . in particular , we obtain a suppression of @xmath181 in the limit of strong detuning , @xmath182 .
we have studied the spin - dependent electronic transport through magnetic molecules for strong coupling to the leads .
our discussion has focused on two complementary regimes . for the first regime , we have presented a description of transport in terms of a master equation that keeps the electronic tunneling exactly , holds for arbitrary bias and gate voltages , and treats the local exchange interaction @xmath3 perturbatively at second order .
this approach is thus applicable for small @xmath3 .
we have derived the bias - dependent magnetic transition rates showing that the tunneling current can be used to drive the molecular spin out of equilibrium .
further , we have shown that the characteristic fine structure of the differential - conductance peaks persists for strong molecule - lead coupling , where the broadening of the peaks is determined by the hybridization energies .
the perturbative expansion in @xmath3 fails if kondo correlations contribute significantly to the transport . in this case , prefactors of the third- and higher - order terms in @xmath3 diverge for @xmath151 .
the kondo correlations can become important for small bias voltages on the order of @xmath183 . here
, transport is described by the meir - wingreen formula in combination with a diagrammatic calculation of the local electronic spectral function of the molecule . we have assumed the molecular level to be far from resonance , which on the one hand makes sure that the resonant - tunneling contributions to the conductance are small and which on the other allows us to obtain analytical results .
we have shown that kondo peaks appear at finite bias voltages proportional to the anisotropy energy of the molecular spin .
our results leave several avenues for future research . first
, it would be interesting to include a local coulomb interaction @xmath184 between the electrons on the molecule .
however , due to the large hybridization there are no states with large probability on the dot and the effect of @xmath184 is expected to be relatively weak .
we expect that for very large @xmath184 an equilibrium kondo resonance could occur as a zero - bias peak in the differential conductance in addition to the nonequilibrium kondo effect described in this paper .
second , the presence of an external magnetic field might lead to an interesting interplay with the splitting of the kondo peaks due to the magnetic anisotropy .
finally , it would be desirable to combine the two cases studied here and to analyze the kondo effect in magnetic molecules in the resonant - tunneling regime , where resonant - tunneling contributions to the conductance are not negligible and the spin is driven out of equilibrium by the current .
we would like to thank a. donabidowicz - kolkowska , d. r. reichman , and a.j .
millis for useful discussions .
financial support by the deutsche forschungsgemeinschaft is gratefully acknowledged .
in this appendix we give details on the derivation of eqs .
( [ currentj ] ) and ( [ currentiaaa ] ) .
we start from eq .
( [ currentj0 ] ) , @xmath185 , h_j(t '' ) \right ] \rho(t).\end{aligned}\ ] ] inserting the expressions for the current operator @xmath186 , eq .
( [ currenti ] ) , and for the exchange interaction @xmath35 , eq .
( [ h_j ] ) , we find @xmath187 , \nonumber \\ & & \quad c^\dagger_{5}(t '' ) c_{6}(t '' ) \mbox{\boldmath$\sigma$}_{\sigma_5\sigma_6}\cdot\mathbf{s}(t '' ) \bigg ] \rho(t ) , \hspace{8em}\end{aligned}\ ] ] where we have assumed @xmath188 to be real .
here , the shorthand notation @xmath189 stands for @xmath190 . introducing @xmath191 and @xmath192 and assuming a product state gives @xmath193 \nonumber \\ & & \qquad{}\times \text{tr}_j \ , 2\ , \mathbf{s}(0)\cdot\mathbf{s}(-\tau ' ) \ , \rho_j \nonumber \\ & & \quad{}- \big [ \delta_{23}\delta_{45}\delta_{16 } f_2 \left ( 1 - f_4 \right ) f_6 e^{i(\epsilon_6-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_4)\tau'/\hbar } - \delta_{14}\delta_{25}\delta_{36 } \left ( 1 - f_2 \right ) \left ( 1 - f_4 \right ) f_6 e^{i(\epsilon_4-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_2)\tau'/\hbar } \nonumber \\ & & \qquad{}- \delta_{14}\delta_{25}\delta_{36 } \left ( 1 - f_2 \right ) f_4 f_6 e^{i(\epsilon_4-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_2)\tau'/\hbar } + \delta_{23}\delta_{45}\delta_{16 } \left ( 1 - f_2 \right ) \left ( 1 - f_4 \right ) f_6 e^{i(\epsilon_6-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_4)\tau'/\hbar } \big ] \nonumber \\ & & \qquad{}\times \text{tr}_j \ , 2\ , \mathbf{s}(0)\cdot\mathbf{s}(\tau ' ) \ , \rho_j \big\}.\end{aligned}\ ] ] this result
can be rewritten as @xmath194 \text{tr}_j \ , 2\ , \mathbf{s}(0)\cdot\mathbf{s}(-\tau ' ) \ , \rho_j \nonumber \\ & & \!\!{}-
\big [ \delta_{23}\delta_{45}\delta_{16 } \left ( 1 - f_4 \right ) f_6 e^{i(\epsilon_6-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_4)\tau'/\hbar } - \delta_{14}\delta_{25}\delta_{36 } \left ( 1 - f_2 \right ) f_6 e^{i(\epsilon_4-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_2)\tau'/\hbar } \big ] \text{tr}_j \ , 2\ , \mathbf{s}(0)\cdot\mathbf{s}(\tau ' ) \ , \rho_j \big\}. \nonumber \\ & & { } \end{aligned}\ ] ] the sums over spin indices are simplified by making use of the identities @xmath195 in the coefficients @xmath70 in eq .
( [ transformation3 ] ) , we approximate the self - energy part by a constant , as we did in sec .
[ sectionmasterequation ] , @xmath196 noting that eqs .
( [ transformation2 ] ) and ( [ transformation3 ] ) imply @xmath197 and @xmath198 we arrive at the following expression for the tunneling current : @xmath199 } \nonumber \\ & & \!\!{}\times \big\ { \big [ \delta_{23}\delta_{45}\delta_{16 } f_4 \left ( 1 - f_6 \right ) e^{i(\epsilon_6-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_4)\tau'/\hbar } - \delta_{14}\delta_{25}\delta_{36 } f_2 \left ( 1 - f_6 \right ) e^{i(\epsilon_4-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_2)\tau'/\hbar } \big ] \text{tr}_j \ , 2\ , \mathbf{s}(0)\cdot\mathbf{s}(-\tau ' ) \ , \rho_j \nonumber \\ & & \!\!{}-\big [ \delta_{23}\delta_{45}\delta_{16 } \left ( 1 - f_4 \right ) f_6 e^{i(\epsilon_6-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_4)\tau'/\hbar } - \delta_{14}\delta_{25}\delta_{36 } \left ( 1 - f_2 \right ) f_6 e^{i(\epsilon_4-\epsilon_2)\tau/\hbar } e^{i(\epsilon_6-\epsilon_2)\tau'/\hbar } \big ] \text{tr}_j \ , 2\ , \mathbf{s}(0)\cdot\mathbf{s}(\tau ' ) \ , \rho_j
\big\}. \nonumber \\\end{aligned}\ ] ] since we have assumed @xmath54 to be diagonal in the stationary state , we finally obtain eqs .
( [ currentj ] ) and ( [ currentiaaa ] ) .
in order to use the meir - wingreen formula for the conductance , we have to compute the imaginary part of the green s function in eq .
( [ eq.img.2 ] ) .
we consider the situation of strongly asymmetric molecule - lead couplings and capacitances , i.e. , @xmath114 , @xmath115 , and @xmath116 , @xmath200
. since wick s theorem does not apply to spin operators , averages of products of spin operators do not factorize into averages of pairs .
we follow ref . in evaluating the spin averages . expanding the electronic matsubara - green s function in powers of @xmath3 and organizing the expansion in terms of topologically distinct diagrams ,
one obtains@xcite @xmath201 \big\rangle_0 \nonumber \\ & ~~\times \bigg\langle t_\tau \left [ b^{\dagger}_{\sigma_1}(\tau_1 ) \frac{\mbox{\boldmath$\sigma$}_{\sigma_1\sigma'_1}}{2 } b_{\sigma_1'}(\tau_1 ) \cdots b^{\dagger}_{\sigma_n}(\tau_n ) \frac{\mbox{\boldmath$\sigma$}_{\sigma_n\sigma'_n}}{2 } b_{\sigma_n'}(\tau_n ) c_{\alpha\textbf{k}\sigma}(\tau ) c^\dagger_{\alpha'\textbf{k}'\sigma'}(\tau ' ) \right ] \bigg\rangle_0,\end{aligned}\ ] ] where @xmath202 denotes the inverse thermal energy . for convenience ,
we have defined @xmath203 .
all non - vanishing diagrams up to third order in @xmath3 are shown in fig .
[ diagrams ] .
the linear term vanishes , since @xmath204 .
diagrams with fermion loops are zero for the following reasons:@xcite a loop with a single fermion line results in taking the trace of the pauli matrix in the vertex , which yields zero .
a loop with two fermion lines appearing in the third - order diagrams gives rise to a trace over two pauli matrices , @xmath205 .
the resulting spin average @xmath206 \rangle_0 $ ] , with at least two of @xmath207 , @xmath208 , and @xmath209 equal , vanishes .
splitting off the zero - order term , the green s function in eq .
( [ appb.green ] ) can be written as@xcite @xmath210 where the unperturbed matsubara - green s function in the imaginary - time domain is given by @xmath211 e^{-\omega_{\alpha\textbf{k}}(\tau-\tau')/\hbar}\ ] ] with @xmath212 . in the frequency domain
we have @xmath213 where @xmath214 is a fermionic matsubara frequency .
note that we are only interested in the spin trace of the self - energy , @xmath215 , which enters in the meir - wingreen formula . the second - order term of the self - energy yields @xmath216 where we have used that @xmath217 .
here , @xmath218 denotes the magnetic anisotropy energy in the spin state @xmath60 with occupation probability @xmath64 .
the spin averages in eq .
( [ appb.green ] ) are to be evaluated for the unperturbed hamiltonian @xmath219,@xcite leading to @xmath220 .
we restrict ourselves to the off - resonance situation , i.e. , the dark region in fig .
[ fig_2dplots](c ) with negligible resonant - tunneling differential conductance , where the spin essentially remains in equilibrium .
this is certainly satisfied if @xmath141 is large compared to the energy scales relevant for the kondo contributions , @xmath142 and @xmath143
. equation ( [ appb.sigma2.2 ] ) contains a sum over leads , @xmath221 , and a factor of @xmath222 under the sum .
since we have assumed strongly asymmetric couplings , @xmath114 , the sum is dominated by the contribution from the right lead , @xmath223 .
dropping the term with @xmath224 , we note that the green s function @xmath225 in eq .
( [ appb.green.5 ] ) only contains the fermi distribution function for the right lead , which is @xmath226 , since @xmath227 .
importantly , the resulting expression is independent of the bias voltage .
furthermore , we see that eq .
( [ appb.sigma2.2 ] ) contains a factor @xmath228 . from eq .
( [ eq.img.2 ] ) we obtain the same factor so that the contribution from leads @xmath14 , @xmath229 is proportional to @xmath230 .
since we have assumed latexmath:[$|t_l| \ll taking the fourier transform of eq .
( [ appb.green.4 ] ) and performing the analytic continuation , we obtain the retarded green s function @xmath232 \ , \sigma^\text{ret}_{rr\mathbf{k}\mathbf{k}'\sigma\sigma}(\omega ) \nonumber \\ & & \quad{}\times \left[p\,\frac{1}{\omega-\epsilon_{r\mathbf{k } ' } } - i\pi\,\delta(\omega-\epsilon_{r\mathbf{k}'})\right ] , \label{eq.gret.omega.3}\end{aligned}\ ] ] where @xmath233 denotes the principal value .
we assume @xmath141 to be large not only compared to @xmath142 and @xmath143 but also to @xmath100 .
one can then show that the delta - function terms are negligible compared to the principal value terms . including the factors of @xmath234 , @xmath235
, we obtain expressions of the form @xmath236 for the imaginary part of the green s function in eq .
( [ eq.img.2 ] ) we then only require the imaginary part of @xmath237 in eq.([eq.gret.omega.3 ] ) .
taking the imaginary part of the fourier transform of eq .
( [ appb.sigma2.2 ] ) we obtain @xmath238 where we assume constant densities of states for the leads , @xmath239 , and an energy band ranging from @xmath240 to @xmath144 , where @xmath144 is the largest energy scale in our model . the third - order term gives @xmath241 \big\rangle_0 \frac{\sigma_{\sigma\sigma_1}^i}{2 } \frac{\sigma_{\sigma_1\sigma_2}^j}{2 } \frac{\sigma_{\sigma_2\sigma}^k}{2}.\end{aligned}\ ] ] here , the average involving spin operators depends on the time arguments @xmath242 , @xmath243 and @xmath244 , since @xmath245 , @xmath246 and @xmath247 can be different . however , since the self - energy only depends on the differences @xmath248 and @xmath249 , we may set @xmath250 and distinguish the two possibilities @xmath251 and @xmath252 . using that @xmath253 = 2i \ , \epsilon_{ijk}$ ] ,
inserting @xmath254 e^{-\epsilon_{r\textbf{k}}(\tau_1-\tau_3)}$ ] and @xmath255 e^{-\epsilon_{r\textbf{k}}\tau_3}$ ] , and evaluating the integral over @xmath244 , we obtain for @xmath256 @xmath257 e^{-\epsilon_{r\mathbf{k}_2}\tau_1/\hbar } e^{(e_m - e_l)\tau_1/\hbar } \nonumber \\ & & \quad{}- \frac{1-f(\epsilon_{r\mathbf{k}_2})}{\epsilon_{r\mathbf{k}_1}-\epsilon_{r\mathbf { k}_2}+e_n - e_l } \left[1-f(\epsilon_{r\mathbf{k}_1})\right ] e^{-\epsilon_{r\mathbf{k}_1}\tau_1/\hbar } e^{(e_m - e_n)\tau_1/\hbar } \nonumber \\ & & \quad{}- \frac{f(\epsilon_{r\mathbf{k}_1 } ) } { \epsilon_{r\mathbf{k}_1}-\epsilon_{r\mathbf{k}_2}+e_m - e_n } \left[1-f(\epsilon_{r\mathbf{k}_2})\right ] e^{-\epsilon_{r\mathbf{k}_2}\tau_1/\hbar } e^{(e_m - e_l)\tau_1/\hbar } \nonumber \\ & & \quad{}+ \frac{f(\epsilon_{r\mathbf{k}_2})e^{\beta(e_m - e_n)}}{\epsilon_{r\mathbf{k}_1 } -\epsilon_{r\mathbf{k}_2}+e_m - e_n } \left[1-f(\epsilon_{r\mathbf{k}_1})\right ] e^{-\epsilon_{r\mathbf{k}_1}\tau_1/\hbar } e^{(e_n - e_l)\tau_1/\hbar } \bigg\ } .\end{aligned}\ ] ] with @xmath258 under the sum over @xmath245 , @xmath246 , @xmath247 , eq . ( [ timeintegral ] )
simplifies to @xmath259 e^{-\epsilon_{r\mathbf{k}_2}\tau_1/\hbar } e^{(e_m - e_l)\tau_1/\hbar } \nonumber \\ & & \quad{}- \frac{f(\epsilon_{r\mathbf{k}_1 } ) } { \epsilon_{r\mathbf{k}_1}-\epsilon_{r\mathbf{k}_2}+e_m - e_n } \left[1-f(\epsilon_{r\mathbf{k}_2})\right ] e^{-\epsilon_{r\mathbf{k}_2}\tau_1/\hbar } e^{(e_m - e_l)\tau_1/\hbar } \bigg\ } .\end{aligned}\ ] ] computing the fourier transform yields @xmath260 finally , the sum over @xmath263 leads to eq .
( [ selfenergy2nd3rd ] ) for the self - energy . here
we assume @xmath264 for all states @xmath265 and only keep the terms that diverge at @xmath266 and low temperatures .
y. xue and m. a. ratner , phys .
b * 68 * , 115406 ( 2003 ) ; * 68 * , 115407 ( 2003 ) ; in _ nanotechnology : science and computation _ , edited by j. chen , n. jonoska , and g. rozenberg ( springer , berlin , 2006 ) , p. 215 . | transport through molecular magnets is studied in the regime of strong coupling to the leads .
we consider a resonant - tunneling model where the electron spin in a quantum dot or molecule is coupled to an additional local , anisotropic spin via exchange interaction .
the two opposite regimes dominated by resonant tunneling and by kondo transport , respectively , are considered . in the resonant - tunneling regime , the stationary state of the impurity spin
is calculated for arbitrarily strong molecule - lead coupling using a master - equation approach , which treats the exchange interaction perturbatively .
we find that the characteristic fine structure in the differential conductance persists even if the hybridization energy exceeds thermal energies .
transport in the kondo regime is studied within a diagrammatic approach .
we show that magnetic anisotropy gives rise to the appearance of two kondo peaks at nonzero bias voltages . |
surgery ( or cut - and - paste topology ) is a method of modifying a manifold to another one .
surgery was successfully used in the sixties to geometrically realize algebraic invariants of manifolds and leads , for example , to a classification of high dimensional manifolds of a fixed homotopy type .
surgery has also been fruitfully applied to the case of embedding questions , most notably to codimension 2 embeddings , i.e. , knot theory . for an excellent survey ,
see @xcite .
this was pioneered by levine , who used surgery to geometrically realize known knot ( and link ) invariants , such as the alexander polynomial , the alexander module , and well - known concordance invariants of knots , @xcite .
rolfsen used surgery for similar reasons in his reader - friendly introduction to knot theory , @xcite . the key idea behind this is the fact that knots ( or rather , knot projections ) can be unknotted via a sequence of crossing changes , and that a crossing change can be achieved by surgery on a @xmath1-framed unknot as follows : @xmath2 thus , every knot @xmath3 in @xmath0 can be obtained by surgery on a framed link @xmath4 in the complement @xmath5 of an unknot @xmath6 .
we will call such a link @xmath4 , an _ untying link _ for @xmath3 . observe that untying links are framed , and null homotopic in @xmath5 , the interior of a solid torus .
further , their linking matrix is invertible over @xmath7 , since surgery on them gives rise to a . in the above mentioned applications of surgery theory to knot theory ,
one starts with a knot invariant ( such as the alexander polynomial ) with a known behavior under surgery . using this information
, one can realize all values of the invariant by choosing suitable untying links on @xmath5 and performing surgery on them .
this is exactly what levine , rolfsen and others did .
there is one point , though , that the above literature does not usually discuss : a knot can be untied by many different links in @xmath0 .
this is due to the fact that surgery is related to handlebodies that have different descriptions . if two framed links are related by a @xmath8 ( @xmath9 or @xmath10 ) move , then surgery on them gives rise to diffeomorphic manifolds . here
, @xmath11 corresponds to adding a parallel of a link component to another , and @xmath10 corresponds to adding / removing a unit - framed unlink away from a link .
the above discussion applies in the case of knots @xmath3 in _ s _ @xmath12 ( that is , closed 3-manifolds with the same @xmath7-homology as @xmath0 ) , and leads to the following _ surgery map _ : @xmath13 where @xmath14 is the set of knots in homology spheres , and @xmath15 denotes the set of null homotopic links @xmath4 with @xmath7-invertible linking matrix in the complement of a standard unknot @xmath6 in @xmath0
. the next theorem explains what we mean by the _ surgery view of knots_. the surgery map is 1 - 1 and onto .
the technique of proving this theorem is well - known in the seventies ; in a sense it is a relative version of kirby s theorem for manifolds with boundary .
the reader may ask why did the classical literature not state the above result , and why is it of any use nowadays ?
as we mentioned already , if one is interested in _ realizing _ values of known knot invariants , one needs to know that the surgery map is onto .
if one is interested in _ constructing _ a knot invariant , starting from an invariant of null homotopic links in the complement of an unknot , then one needs to know the kernel of the surgery map .
this is precisely the situation that we are interested in . in forthcoming work
@xcite we will use the above theorem in order to construct a new knot invariant , namely a rational version @xmath16 of the kontsevich integral of a knot .
this work motivated us to formulate theorem [ thm.surgeryk ] . in a sense ,
theorem [ thm.surgeryk ] defines knot theory in terms of link theory in a solid torus .
one might think that this is a step in the wrong direction . however , this view of knots is closely related to the _ homology surgery _ view of knots of cappell - shaneson , @xcite .
cappell and shaneson study the closed 3-manifold obtained by @xmath17 surgery on a knot .
this manifold has the homology of a circle but not the @xmath7-equivariant homology of a circle . in @xcite
, we actually construct a rational invariant of @xmath18-links .
boundary links ( @xmath19-links ) , and their cousins , @xmath18-links are a generalization of knots which we now explain .
recall first that every knot @xmath3 in a homology sphere @xmath12 gives rise to a map @xmath20 which maps a meridian to a generator of @xmath7 .
an @xmath18-_link _ is a triple @xmath21 of a link @xmath22 in a homology sphere @xmath12 , where @xmath23 is a map that sends meridians of @xmath22 to generators of the free group @xmath18 .
here , @xmath18 denotes the free group in generators @xmath24 , where @xmath25 is the number of components of @xmath22 .
an @xmath18-link gives rise to a map @xmath26 and the preimage of generic points on each circle gives rise to a disjoint union of surfaces @xmath27 with boundary the @xmath28th component @xmath29 of @xmath22 .
thus , @xmath30 is a _
boundary link_. it turns out that every boundary link has an @xmath18-structure , however not a canonical one . for a discussion of this ,
we refer the reader to @xcite and @xcite .
generalizing theorem [ thm.surgeryk ] , let @xmath6 denote a standard unlink of @xmath25 components in @xmath0 , with a canonical isomorphism @xmath31 . if @xmath4 is a null homotopic link in @xmath32 with @xmath7 invertible linking matrix , then after surgery it gives rise to an @xmath18-link @xmath21 .
the next theorem describes the _ surgery view of @xmath18-links_. the surgery map gives a 1 - 1 onto correspondence @xmath33 as we mentioned above , we will use this theorem in @xcite in order to construct a rational version of the kontsevich integral . at present
, we will use the above theorem to define a noncommutative version of the alexander polynomial of an @xmath18-link , as well as a matrix - valued invariant of @xmath18-links which is essentially equivalent to the blanchfield pairing of their free cover . consider a link @xmath34 . since @xmath35 is null homotopic , it lifts to a link @xmath36 in the free @xmath18-cover @xmath37 of @xmath38 . since @xmath37 is contractible , it follows that we can define linking numbers between the components of @xmath36 , and thus we can define the _ equivariant linking matrix _ @xmath39 of @xmath4 by : @xmath40 where @xmath41 denotes a lift of a component @xmath42 of @xmath4 .
the above sums are finite and @xmath39 is a matrix over the group - ring @xmath43 $ ] of @xmath18 .
the size of the matrix is the number of components of @xmath4 .
the equivariant linking matrix of @xmath4 is a _ hermitian matrix _ ( that is , a matrix @xmath44 over @xmath45 is that satisfies @xmath46 where @xmath47 is the transpose of @xmath44 followed by the natural involution @xmath48 ) , defined up to a minor indeterminacy that depends on the choice of lifts @xmath41 of the components of @xmath4 .
a different choice of lifts replaces the linking matrix @xmath39 by @xmath49 , where @xmath50 is a diagonal matrix with elements of @xmath51 in the diagonal . taking into account the kirby moves
, we can define a map @xmath52 where @xmath53 and where @xmath54 denotes the set of _ simple stable congruence classes of hermitian matrices over @xmath45 , invertible over @xmath7 _
, we mean that the image of a matrix which is defined over @xmath45 under the map @xmath55 is invertible over @xmath7 . ] . following @xcite , we call two hermitian matrices @xmath56 _ simply stably congruent _
iff @xmath57 for some diagonal matrices @xmath58 with @xmath1 entries and some matrix @xmath50 which is either _ elementary _
( i.e. , it differs from the identity matrix on a single non - diagonal entry ) or _ diagonal _ ( with entries in @xmath51 ) .
let @xmath59 denote the completion of the rational group - ring @xmath60 $ ] with respect to the augmentation ideal , that is the ideal generated by @xmath61 , where @xmath24 generate @xmath18 .
@xmath59 can be identified with the ring @xmath62 $ ] of formal power series in non - commuting variables @xmath63 .
let @xmath64 be the invariant on the set @xmath65 _ of matrices @xmath66 over @xmath59 , nonsingular over @xmath7 _ , defined by @xmath67 where @xmath68 and @xmath69 . in a sense , @xmath70 is a logarithmic determinant . in our case , we want to define @xmath70 on the quotient @xmath71 of the set of hermitian matrices over @xmath45 which are invertible over @xmath7 .
unfortunately , @xmath70 does not descend to a @xmath59-valued invariant on the set @xmath71 , but it does descend to a @xmath72-valued invariant , where @xmath72 is the abelian group quotient of the abelian group @xmath59 modulo the subgroup @xmath73 for @xmath74 .
equivalently , @xmath72 can be described as the free abelian group generated by cyclic words in @xmath63 variables .
our next result shows that the @xmath70 invariant is a noncommutative analogue of the alexander polynomial of an @xmath18-link . as a motivation ,
recall that the alexander polynomial of a knot measures the order of the first homology of the free abelian cover of its complement .
given an @xmath18-link @xmath21 , we can consider the free cover @xmath75 of its complement @xmath76 , and its homology @xmath77 , which is a module over the group - ring @xmath45 .
since the ring @xmath45 is not commutative ( when @xmath22 is not a knot ) , the notion of a torsion module does not make sense .
however , in @xcite , levine and the first author defined an invariant @xmath78 of @xmath18-links with values in @xmath79 , the abelian group of formal power series ( with rational coefficients ) in noncommutative variables , modulo cyclic permutation .
the invariant @xmath78 is a reformulation of an invariant defined earlier by farber , @xcite , and in the case of knots equals , up to a multiple , to the logarithm of the alexander polynomial .
thus , @xmath78 can be thought as a _ noncommutative torsion _
invariant of @xmath18-links .
the next result gives an independent definition of the noncommutative torsion of an @xmath18-link , and of its blanchfield form .
there exist an invariant @xmath66 of @xmath18-links with values in @xmath54 .
@xmath77 is a @xmath45-module with presentation @xmath66 .
the blanchfield pairing @xcite @xmath80 is given by @xmath81 , where @xmath82 is the cohn localization of @xmath45 , @xcite .
@xmath83 .
thus , the @xmath66 invariant of an @xmath18-link determines its blanchfield form .
conversely , using various exact sequences of algebraic surgery , one can show that the blanchfield form determines @xmath66 , as was communicated to us by d. sheiham .
since this would take us far afield , we prefer to postpone this to a future publication .
fix once and for all a _ based unlink _ @xmath6 of @xmath84 components in @xmath0 .
a based unlink consists of an unlink @xmath6 in @xmath0 , together with arcs from a base point in @xmath85 to each of the component of the unlink , as shown below ( for @xmath86 ) : @xmath87 let @xmath15 denote the set of _ null homotopic links @xmath22 with @xmath7-invertible linking matrix _ in the complement of @xmath6 .
surgery on an element @xmath4 of @xmath15 transforms @xmath88 to a @xmath18-link @xmath30 . indeed , since @xmath4 is null homotopic the natural map @xmath89 gives rise to a map @xmath90 .
alternatively , one can construct disjoint seifert surfaces for each component of @xmath22 by tubing the disjoint disks that @xmath6 bounds , which is possible , since each component of @xmath4 is null homotopic . since the linking matrix of @xmath4 is invertible over @xmath7
, @xmath12 is an .
let @xmath91 denote the equivalence relation on @xmath15 generated by the moves of _ handle - slide _ @xmath11 ( i.e. , adding a parallel of a link component to another component ) and _ stabilization _
@xmath10 ( i.e. , adding to a link an unknot away from the link with framing @xmath1 ) .
it is well - known that @xmath8-equivalence preserves surgery , thus we get a surgery map @xmath92 now , we will show that the surgery map is onto . given a @xmath18-link @xmath21
choose disjoint seifert surfaces @xmath93 for its components and choose a framed link @xmath4 in @xmath12 such that @xmath94 .
@xmath4 might intersect @xmath93 , however by a small isotopy ( which does not change the property that @xmath94 ) we may assume that @xmath4 is disjoint from @xmath93 .
consider the image @xmath95 of @xmath93 in @xmath96 . by reversing the @xmath4 surgery on @xmath12 ,
consider the image @xmath97 of a parallel of @xmath4 in @xmath98 such that @xmath99 . by doing crossing changes among the bands of @xmath95 , translated in terms of surgery on a unit - framed link @xmath100
, we can assume that each component of @xmath95 lies in a ball disjoint from the other components and further that the boundary of @xmath95 is an unlink .
the link @xmath101 is null homotopic in the complement of the unlink @xmath102 since it is disjoint from @xmath95 , and since the map @xmath103 is given geometrically by intersecting a loop in @xmath5 with @xmath95 and recording the corresponding element of the free group .
the linking matrix of @xmath101 is invertible over @xmath7 since @xmath104 , an .
now perform an isotopy and choose arcs @xmath105 from a base point to each of the components of @xmath95 in such a way that @xmath106 intersects a ball as follows ( for @xmath86 ) : @xmath107 by construction , @xmath108 maps to @xmath21 .
the goal of this independent section is to show a relative version of fenn - rourke s theorem ( * ? ? ?
* theorem 6 ) .
we begin with some notation . given a 3-manifold @xmath12 ( possibly with nonempty boundary ) and a framed link @xmath22 in its interior ,
surgery produces a 3-manifold @xmath109 whose boundary is canonically identified with that of @xmath12 and a cobordism @xmath110 with boundary @xmath111 . *
@xmath114 , a diffeo rel . boundary . * an isomorphism @xmath115 . * a commutative diagram @xmath116 @xmath117{e , t}{h_\ast}\arrow{s } \node[2]{\pi_1(m_{l_2 } ) } \arrow{s } \\ \node{\pi_1(w_{l_1 } ) } \arrow[2]{e , t, .. }{\iota } \node[2]{\pi_1(w_{l_2 } ) } \\
\node[2]{\pi_1(m ) } \arrow{nw}\arrow{ne } \end{diagram}\ ] ] here , @xmath119 is the homology class of @xmath120 , where @xmath121 , @xmath122 ( where the boundaries are identified via the identity on @xmath12 , @xmath123 and @xmath124 on the top piece ) and @xmath125 .
note that @xmath66 is a closed 4-manifold .
the proof of theorem [ thm.fr ] follows mainly from fenn - rourke @xcite , with some simplifications due to roberts , @xcite .
we begin with the following lemma from @xcite , which reformulates the vanishing of the condition @xmath126 .
* lemma 9 ) assume that we are in the situation of theorem [ thm.fr ] and that there exists @xmath127 making @xmath128 commute .
then @xmath126 iff @xmath129 for some @xmath130 and closed 5-manifold @xmath131 such that the diagram @xmath132{e , t}{\iota}\arrow{se , r}{(j_1)_\ast } \node[2]{\pi_1(w_{l_2 } ) } \arrow{sw , r}{(j_2)_\ast } \\
\node[2]{\pi_1(\omega^5 ) } \end{diagram}\ ] ] commutes and @xmath133 are split injections .
( of theorem [ thm.fr ] ) if @xmath112 and @xmath113 are @xmath8-equivalent , then it follows from the `` only if '' statement of lemma [ lem.fr ] that @xmath126 since the induced homeomorphisms on @xmath134 do not change .
the next step is to use cerf theory .
namely , consider morse functions @xmath138 realizing the cobordisms @xmath139 for @xmath140 .
since @xmath135 is diffeomorphic to @xmath136 , @xmath141 and @xmath142 are related by a one parameter family of smooth functions @xmath143 for @xmath144 .
the arguments of roberts ( * ? ? ?
* stage 2 , p.3 ) in this case imply that @xmath112 is @xmath8-equivalent to @xmath113 .
_ end of the proof of theorem [ thm.surgeryf ] .
_ we will apply theorem [ thm.fr ] to @xmath147 .
consider @xmath148 two surgery presentations of the same @xmath18-link for @xmath140 .
then , we have a homotopy commutative diagram @xmath149 where @xmath150 are splitting maps .
it follows by remark [ rem.fr ] that a map @xmath151 exists such that the following diagram @xmath152 commutes @xmath153{e , t}{h_\ast}\arrow{s}\arrow[2]{se , l,1}{\phi_1 } \node[4]{\pi_1(m_{l_2 } ) } \arrow{s}\arrow[2]{sw , l,1}{\phi_2 } \\ \node{\pi_1(w_{l_1 } ) } \arrow{e,- , .. } \node { } \arrow[2]{e , t,-, .. }{\iota } \node[2 ] { } \arrow{e , .. } \node{\pi_1(w_{l_2 } ) } \\
\node[3]{\pi_1(m ) } \arrow{nww , r}{\sim}\arrow{nee , r}{\sim } \end{diagram}\ ] ] the obstruction @xmath154 lies in @xmath155 .
however , @xmath156 , thus @xmath157 .
thus , theorem [ thm.fr ] implies that @xmath112 is @xmath8-equivalent to @xmath113 .
the goal of this section is to prove theorem [ thm.classical ] .
combining the map @xmath158 of definition [ def.w ] with the surgery map of theorem [ thm.surgeryf ] , gives rise to the map @xmath66 of theorem [ thm.classical ] ( a ) . for matrices
@xmath56 over @xmath59 , nonsingular over @xmath7 , we have in @xmath72 that @xmath159 ( b ) @xmath70 descends to a @xmath72-valued invariant of the set @xmath54 .
for @xmath44 as above , @xmath160 where @xmath69 . if @xmath44 is a matrix with integer entries invertible over @xmath7 , then @xmath161 .
\(b ) follows from ( a ) . for ( a ) ,
let @xmath162 denote the ring of square matrices with entries in @xmath59 , and @xmath163 $ ] denote the subgroup of it generated by matrices of the form @xmath164 where @xmath165 . given two matrices @xmath165 , with augmentations @xmath166 and @xmath167
, we claim that @xmath168.\ ] ] indeed , if @xmath169 , then this follows from the baker - cambell - hausdorff formula , since @xmath170 . for the general case ,
we have modulo @xmath163 $ ] , that @xmath171 since @xmath172 after taking traces , equation implies ( a ) .
follows immediately from the definition of @xmath70 and the fact that @xmath173
. follows from that fact that if @xmath44 is invertible over @xmath7 , then @xmath174 , which implies @xmath175 , thus @xmath176 .
( of theorem [ thm.classical ] ) consider @xmath34 and let @xmath21 denote the corresponding @xmath18-link .
let @xmath66 denote the equivariant linking matrix of @xmath4 , and @xmath75 ( resp .
@xmath37 ) denote the free cover of @xmath76 ( resp .
@xmath178 ) given by @xmath179 . note that @xmath75 is obtained from @xmath37 by surgery on @xmath36 .
part ( b ) follows from classical surgery arguments presented in the case of knots by levine in @xcite and adapted without difficulty in the case of @xmath18-links ; see also ( * ? ? ?
* lemma 2.4 ) . for the case of knots , lemma 12.2 and the discussion of p.46 - 47 in @xcite ,
imply that the blanchfield pairing is given by @xmath180 modulo @xmath45 .
this discussion can be generalized without change to the case of @xmath18-links , and proves part ( c ) . for part ( d ) , observe that the @xmath70 and the @xmath78 invariant of @xcite are defined , respectively , via the seifert surface and the surgery view of boundary links .
the two views can be related using the so - called y - view of knots presented in ( * ? ? ?
* section 6.4 ) and extended without difficulty to the case of boundary links . consider a disjoint union @xmath93 of seifert surfaces in @xmath0 which we think of as a disjoint union of embedded disks with pairs of bands attached in an alternating way along each disk .
consider an additional link @xmath181 in @xmath182 such that its linking matrix @xmath4 satisfies @xmath183 and such that the linking number between the cores of the bands and @xmath181 vanishes .
with respect to a choice of orientation of 1-cycles corresponding to the cores of the bands , a seifert matrix of @xmath93 is given by @xmath184 start from a surgery presentation of @xmath185 in terms of _ clovers with three leaves _ , as was explained in ( * ? ? ?
* section 6.4 ) and summarized in the following figure : @xmath186 surgery on a clover with three leaves can be described in terms of surgery on a six component link @xmath187 .
it was observed by the second author in ( * ? ? ?
* figure 3.1 ) that @xmath187 can be simplified via kirby moves to a four component link @xmath188 disjoint from a surface @xmath95 shown as follows ( where @xmath189 correspond to the same link in the figure above and below ) : @xmath190 the equivariant linking matrix of ( a based representative of ) @xmath191 is given by @xmath192 where @xmath66 is the following matrix in @xmath193 basis : | a celebrated theorem of kirby identifies the set of closed oriented connected 3-manifolds with the set of framed links in @xmath0 modulo two moves . we give a similar description for the set of knots ( and more generally , boundary links ) in homology 3-spheres . as an application
, we define a noncommutative version of the alexander polynomial of a boundary link .
our surgery view of boundary links is a key ingredient in a construction of a rational version of the kontsevich integral , which is described in subsequent work .
stavros .
1991 _ mathematics classification . _ primary 57n10 .
secondary 57m25 . ] |
in the study of totally disconnected locally compact ( t.d.l.c . ) hausdorff groups , groups of automorphisms of locally finite trees appear naturally and form a significant class of examples .
they are also the most basic case of groups acting on buildings or cat(0 ) cube complexes .
moreover , they play an important role in the structure theory of compactly generated t.d.l.c .
groups through schreier graph constructions ( see @xcite and m. burger s chapter ) . in section 2 ,
we present general results and notions about automorphisms of infinite , locally finite trees .
section [ sec : tits ] deals with independence properties ( the idea that restrictions of the action to subtrees are in some sense independent of each other ) and a general form of tits simplicity theorem , theorem [ tits - k ] .
this theorem extracts abstractly simple subgroups from groups acting on the tree in a sufficiently dense and independent way .
we then present two constructions of closed subgroups of tree automorphisms with prescribed local actions and examine some of the local - global phenomena that they exhibit .
first , we present in section [ sec : universal ] the universal group construction by m. burger and s. mozes ( see @xcite ) , which to every permutation group @xmath1 associates a subgroup @xmath2 which locally acts like @xmath3 ( here , @xmath4 denotes the @xmath5-regular tree ) .
properties of the groups @xmath6 are often determined by those of the finite group @xmath3 .
universal groups are fundamental in the study of lattices in the product of the automorphism groups of two trees , analogous to the study of lattices in semisimple lie groups , and are key to the proof of the normal subgroup theorem ( see @xcite and l. bartholdi s chapter ) .
second , we describe in section [ sec : k - closure ] a variation of this construction by c. banks , m. elder and g. willis ( see @xcite ) : given @xmath7 for an arbitrary tree @xmath8 it produces a sequence of closed subgroups @xmath9 which act like @xmath10 on balls of radius @xmath0 .
this sequence converges to the topological closure @xmath11 of @xmath10 from above , in a sense to be made precise .
we will use tits simplicity theorem to construct infinitely many , pairwise distinct , non - discrete , compactly generated , abstractly simple , t.d.l.c . groups . finding simple t.d.l.c .
groups ( and in particular , compactly generated ones ) has become relevant to the structure theory of t.d.l.c .
groups thanks to results of caprace
monod @xcite , which state that such groups can be decomposed into simple pieces .
let @xmath8 be a simplicial tree with vertex set @xmath12 and edge set @xmath13 .
we will always assume that @xmath12 is countable and that @xmath8 is not isomorphic to a bi - infinite line .
each edge @xmath14 is determined by its origin @xmath15 and its terminus @xmath16 , so the edge @xmath17 is the _ inverse _ of @xmath18 .
the pair @xmath19 is a _
geometric edge_. given any edge @xmath20 , the subgraph @xmath21 has two connected components @xmath22 which we call _ half - trees_. we think of @xmath23 as a subtree rooted at @xmath24 . a _ path _
is given by a sequence of adjacent vertices , and it is a _ reduced path _ ( or _ geodesic _ ) if there is no backtracking in the sequence . setting adjacent vertices to have distance 1 between them yields the standard metric on @xmath8 , denoted by @xmath5 .
given a vertex @xmath25 , we set @xmath26 to be the set of edges starting at @xmath24 and write @xmath27 for the ball of radius
@xmath28 centred at @xmath24 as well as @xmath29 for the sphere of radius @xmath28 around @xmath24 .
an isometric embedding of @xmath30 into @xmath8 is called a _ ray_. we will say that two rays @xmath31 are equivalent , denoting it by @xmath32 , if there exists some @xmath33 such that @xmath34 .
we call the collection of equivalence classes of such rays the _ boundary _ of the tree and denote it by @xmath35 the rays @xmath36 are _ at bounded distance from each other _ if the function @xmath37 is bounded .
this _ a priori _ weaker equivalence relation is actually the same as the previous one : it is impossible for two rays to be at bounded distance from each other without actually coinciding eventually .
we introduce a topology on @xmath38 ( resp . on @xmath39 ) by taking @xmath40 ( resp . @xmath41 as a basis of open sets .
it is easy to see that this makes @xmath8 into an discrete , open and dense subset of @xmath39 .
when @xmath8 is locally finite then both @xmath38 and @xmath39 are compact with this topology .
let @xmath42 be the automorphism group of @xmath8 .
suppose that @xmath43 and that @xmath44 is a subgraph of @xmath8 .
the _ stabilizer _
@xmath45 of @xmath44 consists of elements @xmath46 such that @xmath47 , while the _ fixator _ @xmath48 of @xmath44 consists of @xmath46 such that @xmath49 for every vertex @xmath50 of @xmath44 .
we say that @xmath10 acts _ edge - transitively _ ( respectively , _
vertex - transitively _ ) if @xmath10 acts transitively on geometric edges ( respectively , vertices ) .
the topology of pointwise convergence on vertices gives @xmath42 the structure of a polish ( i.e. metrizable , separable and complete ) topological group .
the basic open sets for this topology are of the form @xmath51 where @xmath52 and @xmath53 is a finite subset of @xmath12 . in particular , @xmath54 is open for any finite @xmath53 .
we will sometimes assume that the tree is locally finite ( i.e. every vertex has only finitely many neighbors ) .
in this case the balls @xmath55 around any vertex @xmath24 are finite . since these balls are preserved by the stabilizer @xmath56 , this open subgroup is the inverse limit of its restriction to balls of finite radius and is therefore compact .
thus when @xmath8 is locally finite @xmath42 is a t.d.l.c .
group .
we will say that two directed edges are _ co - oriented _ if their orientations agree along the unique geodesic connecting them ( see figure [ fig : co_oriented ] ) . for an automorphism @xmath57 , define @xmath58 and @xmath59 .
the minimum above is taken over all points @xmath24 in the geometric realization of the tree .
thus , for example , if @xmath60 inverts an edge @xmath18 then the set @xmath61 will be the midpoint of the corresponding geometric edge . due to the simplicial nature of the action
, this minimal set is actually realized and we are allowed to talk about the minimum rather than the infimum of the translation length .
[ def : classes ] an automorphism @xmath60 is called _ hyperbolic _ if @xmath62 , an _ inversion _ if it inverts an edge and _ elliptic _ if it fixes some vertex .
it is clear that @xmath42 is a disjoint union of these three classes and that all three classes , which we denote by @xmath63 , are clopen .
let @xmath57 and assume that there exists an edge @xmath18 that is co - oriented with its image @xmath64 .
since being co - oriented is an equivalence relation on directed edges , it follows that @xmath65 are all co - oriented along a bi - infinite geodesic @xmath66 and @xmath60 restricts to a translation of length @xmath67 along this axis .
once we have such an invariant axis , then for every vertex @xmath68 we have @xmath69 ( see figure [ fig : co_oriented ] ) .
in particular , as our notation already suggests , @xmath70 . thus , whenever there exists an edge that is co - oriented with its image , @xmath60 is hyperbolic .
conversely , if @xmath62 , let @xmath71 , and let @xmath72 be the geodesic connecting @xmath24 to @xmath73 . the directed edge @xmath74 must be co - oriented with its image since otherwise we would have @xmath75 , contradicting our choice of @xmath24 .
a similar picture shows that the general formula @xmath76 holds also for elliptic elements and inversions .
when @xmath77 we denote by @xmath78 the two points of @xmath38 corresponding to the two extremes of the axis @xmath61 .
the point @xmath79 , represented by the ray in the direction of translation , is called _ the attracting point of _ and @xmath80 is called its _ repelling point_. note that if @xmath81 are open neighborhoods , then for every large enough @xmath82 we have @xmath83 .
since @xmath84 and @xmath60 share the same axis , but translate in different directions , we have @xmath85 .
if we replace the tree by its barycentric subdivision ( adding one vertex in the middle of every geometric edge ) , we retain the same automorphism group , but every inversion becomes an elliptic element on the new tree . using this trick
we will assume from now on that there are no inversions in @xmath42 .
a subgroup @xmath86 is _ purely elliptic _
( respectively , _ purely hyperbolic _ ) if all of its ( non - trivial ) elements are elliptic ( respectively , hyperbolic ) .
[ lem : tits ] ( ( * ? ? ?
* proposition 26 ) ) let @xmath87 be two elliptic elements with @xmath88
. then @xmath89 .
let @xmath90 be the shortest geodesic from @xmath61 to @xmath91 .
let @xmath74 be the first directed edge along this path .
it is clear from figure [ fig : tits_lem ] that @xmath18 and @xmath92 are co - oriented which proves the lemma .
the following is a helly type lemma , capturing the fact that a tree is essentially a one - dimensional object .
[ lem : helly ] if @xmath93 are convex subsets of @xmath8 with pairwise non - empty intersections then @xmath94 . we refer to the indices modulo @xmath95 .
pick a point @xmath96 . by convexity ,
the geodesic from @xmath97 to @xmath98 is contained in @xmath99 .
now , in any tree , every triangle has at least one point common to all of its edges ( a tree is @xmath100-hyperbolic ) . for the triangle with vertices @xmath101 this point will be in the desired intersection @xmath102 .
[ cor : pe ] let @xmath86 be a purely elliptic subgroup
. then @xmath10 fixes a point in @xmath39 .
if @xmath10 is finitely generated then it fixes a point in @xmath8 .
assume first that @xmath10 is generated by a finite set @xmath103 .
by lemma [ lem : tits],for any pair @xmath104 of generators , @xmath105 .
now , applying lemma [ lem : helly ] successively , we find that @xmath106 , which proves the finitely generated case .
assume that @xmath10 is purely elliptic but does not fix any point within the tree .
given @xmath107 and @xmath108 such that @xmath109 we let @xmath110 be the neighbour of @xmath24 in the direction of @xmath111 .
the point is that @xmath112 does not depend on @xmath113 as , by the finitely generated case , @xmath114 are convex subtrees with a non - trivial intersection .
consider the geodesic ray @xmath115 .
for two different vertices @xmath116 we have @xmath117 since we can choose some @xmath113 that fixes neither @xmath24 nor @xmath50 and @xmath118 would then just be one step towards the convex set @xmath111 .
thus @xmath119 represent the same point in the boundary .
moreover this boundary point is fixed by @xmath10 as @xmath120 .
[ def : geom_dense ] a group @xmath86 is called _ geometrically dense _ if it does not fix ( pointwise ) any end of @xmath8 and does not stabilize ( as a set ) any proper subtree of @xmath8 .
this notion appears as a condition in tits simplicity theorem ( our theorem [ tits - k ] ) .
it can be generalized to much more general cat(0 ) spaces , and in this capacity plays an important role in the work of caprace
monod @xcite who argue that it should be thought of as a a geometric analogue of zariski density .
for example , given a local field @xmath0 and considering the action of @xmath121 on its bruhat tits tree , a subgroup @xmath122 is geometrically dense if and only if it is zariski dense as a subgroup of @xmath123 .
here @xmath124 is the algebraic closure of @xmath0 .
[ lem : min_tree ] if @xmath86 contains at least one hyperbolic element then there is a unique minimal @xmath10-invariant subtree of @xmath8 . if @xmath125 then its axis @xmath111 is contained in every nonempty convex @xmath10-invariant set ( refer to figure [ fig : co_oriented ] )
let @xmath126 be the smallest convex set containing the axes of all hyperbolic elements of @xmath8 .
since @xmath127 the collection @xmath128 is invariant under @xmath10 and so is its convex hull @xmath44 .
thus @xmath44 is an invariant subtree which is contained in any other @xmath10-invariant tree .
[ rem : densel ] in the setting of the above lemma , let @xmath129 be the collection of attracting points of all the hyperbolic elements in @xmath10 .
as @xmath130 is @xmath10-invariant , so is its convex hull @xmath131 .
thus @xmath132 is a @xmath10-invariant tree .
since @xmath10 contains a hyperbolic element @xmath113 , the set @xmath130 contains at least two points @xmath133 and @xmath44 is non - empty . if we further assume that @xmath10 is geometrically dense then @xmath134 and consequently @xmath130 must be dense in @xmath38 . [
lem : gd_hyp ] every geometrically dense subgroup @xmath135 contains a hyperbolic element .
furthermore , given any half - tree @xmath126 , the group @xmath10 contains a hyperbolic element whose axis is contained in @xmath44 .
recalling that we have assumed away the existence of inversions , the existence of hyperbolic elements follows directly from corollary [ cor : pe ] . by remark [ rem : densel ] above for every half - tree @xmath44 , @xmath136 contains an attracting point @xmath137 for some @xmath138 .
let @xmath139 be another hyperbolic element with attracting and repelling fixed points @xmath140 .
the conjugations @xmath141 are again hyperbolic elements with @xmath142 . if @xmath143 then , due to the proximal action of @xmath144 on @xmath38 , all of these can be `` pushed '' arbitrarily close to @xmath137 , and hence deep into the half - tree @xmath44 , by applying high powers of @xmath144 .
thus it is enough to find such a @xmath113 .
we choose an element of the form @xmath145 where @xmath146 .
this is hyperbolic with attracting and repelling points @xmath147 .
assume by way of contradiction that @xmath148for all @xmath146 .
since @xmath10 is a group this means that either the point @xmath149 or the set @xmath150 is @xmath10-invariant , contradicting our assumptions that @xmath10 is geometrically dense and that the tree is not an infinite line .
[ lem : gd_normal ] let @xmath135 be geometrically dense and assume that @xmath151 is a nontrivial group normalized by @xmath152 .
then @xmath153 is geometrically dense too .
we start with the boundary .
let @xmath154 .
we claim that @xmath155 is empty .
clearly @xmath155 is closed and , since @xmath153 is normalized by @xmath10 , the collection of its fixed points is stabilized by @xmath10 .
if @xmath155 consists of one point then this point is fixed by @xmath10 , contradicting geometric density . if @xmath156 let @xmath44 be the convex hull of @xmath155 inside the tree @xmath8 .
this is a non - empty @xmath10-invariant subtree of @xmath8 and , by the geometric density of @xmath10 , we have @xmath134 .
this readily implies that @xmath155 is dense and hence @xmath157 , contradicting our assumption that @xmath153 is nontrivial .
thus @xmath153 has no fixed point on the boundary .
a very similar argument shows that @xmath153 does not fix any point inside the tree . since @xmath153
does not admit global fixed points in @xmath39 , corollary [ cor : pe ] implies that @xmath153 contains hyperbolic elements .
thus , by lemma [ lem : min_tree ] , there is a unique minimal @xmath153-invariant subtree @xmath126 .
this unique tree is stabilized by @xmath10 . since the latter is geometrically dense we actually have @xmath134 , which proves the theorem
for any finite or ( bi-)infinite path @xmath158 in @xmath8 and any @xmath159 let @xmath160 denote the @xmath0-neighborhood of @xmath158 ( i.e. the subtree of @xmath8 spanned by all vertices at distance at most @xmath0 from @xmath158 ) .
denote by @xmath161 the nearest point projection on @xmath158 and let @xmath162 . if @xmath163 , then , for each vertex @xmath24 of @xmath158 , the pointwise stabilizer @xmath164 acts on @xmath23 . denoting by @xmath165 the permutation group induced by restricting @xmath166 to @xmath23
, we obtain a map @xmath167 which is clearly an injective homomorphism .
[ fig : pk - def ] we say that @xmath10 satisfies _ property @xmath168 _ if for every finite or ( bi-)infinite path @xmath158 in @xmath8 the map @xmath169 defined above is an isomorphism .
notice that when @xmath170 we recover the original property @xmath171 defined by tits ( @xcite ) so we sometimes omit the subscript when referring to property @xmath172 .
we remark that property @xmath171 is also known as tit s independence property in the literature , because it ensures that the actions on subtrees rooted at a path can be chosen independently from each other . to find simple subgroups of @xmath42 we will use a generalization of tits theorem ( ( * ? ?
* thorme 4.5 ) ) .
let @xmath173 denote the subgroup of @xmath10 generated by pointwise stabilizers of `` @xmath174-thick '' edges .
in particular , @xmath175 is generated by pointwise stabilizers of edges .
[ tits - k ] suppose that @xmath176 is geometrically dense and satisfies property @xmath168 .
then @xmath177 is either simple or trivial .
figure [ fig : simplicity ] illustrates the proof .
write @xmath178 .
we assume that @xmath152 is nontrivial as otherwise there is nothing to prove .
let @xmath179 be normalized by @xmath152 .
we will show that @xmath180 , thus proving the theorem .
since @xmath152 is generated by the pointwise stabilizers @xmath181 , it is enough to fix an edge @xmath182 and show that @xmath183 .
let @xmath116 be the vertices incident with @xmath18 .
writing @xmath184 , by property @xmath168 , there is a decomposition @xmath185 . by the symmetry of the situation
it is enough to show that @xmath186 . by lemma [ lem : gd_normal ]
@xmath187 is geometrically dense .
a second application of the same lemma shows that @xmath153 is geometrically dense too .
lemma [ lem : gd_hyp ] shows that there exists a hyperbolic element @xmath188 with an axis whose @xmath0-neighbourhood is completely contained in the half - tree @xmath23 .
we denote the axis by @xmath189 with @xmath190 where @xmath191 is the translation length .
assume that @xmath192 is the projection of the edge @xmath18 on this axis .
we have to show that a given element @xmath193 is contained in @xmath153 .
applying property @xmath168 to the axis @xmath158 we obtain @xmath194 . in the last equality we used the fact that @xmath195 for every @xmath196 which implies that @xmath197 .
since @xmath153 is normalized by @xmath152 , we have @xmath198 = n \psi n^{-1 } \psi^{-1 } \in n$ ] for all @xmath199 . to prove that @xmath200 it is enough to exhibit some @xmath199 such that @xmath201 $ ] .
we will in fact find such an element in @xmath164 . if @xmath202 is an element we denote by @xmath203 its restriction to @xmath204 . by property @xmath168 we have the freedom to construct @xmath205 by prescribing each element @xmath206 separately .
with this notation we have @xmath198_i = ( n \psi n^{-1 } \psi^{-1})_i = n|_{i-\ell } \circ \psi_{i-\ell } \circ ( n^{-1})|_{i } \circ ( \psi^{-1})_i$ ] . solving for @xmath198 = \phi$ ]
we obtain two equations @xmath207 where for the second one we shifted all indices by @xmath208 . now assuming we have arbitrarily fixed the values of @xmath209 we can now solve recursively for all other values of @xmath210 using the first of the above equations for the positive values of @xmath211 and the second one for negative values of @xmath211 .
this method enables us to realize any element @xmath212 as a commutator of the form @xmath213 $ ] .
note that in our specific case we have @xmath197 for all @xmath196 .
this completes the proof .
in this section we introduce the universal group construction that was developed and studied by m. burger and s. mozes in @xcite , and later on played an important role in the study of lattices in the product of the automorphism groups of two regular trees , see @xcite .
let @xmath214 be the @xmath5-regular tree ( @xmath215 ) . recall that @xmath216 denotes the set of edges with origin @xmath217 .
further , let @xmath218 be a legal labelling of @xmath4 , i.e. for every @xmath217 the map @xmath219 is a bijection , and @xmath220 for all @xmath221 .
a ball of radius two in @xmath222 looks as follows : @xmath223(0,0 ) \psline(0,0)(1,0 ) \psline(0,0)(-0.5,0.86 ) \psline(0,0)(-0.5,-0.86 ) \psdots[](1,0 ) \psdots[](-0.5,0.86 ) \psdots[](-0.5,-0.86 ) \uput[u](0.5,0){$1 $ } \uput[30](-0.25,0.43){$2 $ } \uput[150](-0.25,-0.43){$3 $ } \psline(1,0)(1.5,-0.86 ) \psline(1,0)(1.5,0.86 ) \psdots[](1.5,-0.86 ) \psdots[](1.5,0.86 ) \uput[30](1.25,-0.43){$2 $ } \uput[330](1.25,0.43){$3 $ } \psline(-0.5,0.86)(-1.5,0.86 ) \psline(-0.5,0.86)(0,1.73 ) \psdots[](-1.5,0.86 ) \psdots[](0,1.73 ) \uput[150](-0.25,1.30){$3 $ } \uput[u](-1,0.86){$1 $ } \psline(-0.5,-0.86)(-1.5,-0.86 ) \psline(-0.5,-0.86)(0,-1.73 ) \psdots[](-1.5,-0.86 ) \psdots[](0,-1.73 ) \uput[210](-0.25,-1.30){$2 $ } \uput[d](-1,-0.86){$1 $ } \endpspicture\ ] ] now , given @xmath217 , every automorphism @xmath224 induces a permutation at @xmath24 via the following map : @xmath225 [ def : universal_groups ] let @xmath1 .
define @xmath226 .
it is immediate from the following cocycle property of the map @xmath227 that the sets introduced in definition [ def : universal_groups ] are in fact groups .
[ lem : cocycle_identity ] let @xmath217 and @xmath228 . then @xmath229 .
the groups of definition [ def : universal_groups ] are termed _ universal groups _ because of proposition [ prop : universality ] below . in words , they consist of those automorphisms of the regular tree which around every vertex act like one of the allowed permutations from @xmath3 . to determine the dependence of @xmath230 on the labelling @xmath231 , we record the following .
[ lem : ext ] given a quadruple @xmath232 consisting of legal labellings @xmath233 of @xmath4 and vertices @xmath234 , there is a unique automorphism @xmath224 with @xmath235 and @xmath236 . by assumption , @xmath235 .
now assume inductively that @xmath113 is uniquely determined on @xmath237 @xmath238 and let @xmath217 be at distance @xmath28 from @xmath239 .
then @xmath113 is also uniquely determined on @xmath216 by the requirement that @xmath236 , namely @xmath240 .
[ cor : labellings ] let @xmath231 and @xmath241 be legal labellings of @xmath4 .
further , let @xmath1
. then @xmath242 and @xmath243 are conjugate in @xmath244 .
choose @xmath245 . if @xmath246 is the automorphism of @xmath4 associated to @xmath247 by lemma [ lem : ext ] , then @xmath248 .
with corollary [ cor : labellings ] in mind , we henceforth omit the reference to an explicit labelling . [
ex : uf ] clearly , @xmath249 . on the other hand , @xmath250 where the number of copies of @xmath251 is @xmath5 . to see this , fix @xmath245 and for @xmath252 denote by @xmath253 the edge with @xmath254 and @xmath255 .
further , let @xmath256 denote the unique label - respecting inversion of the edge @xmath257 , which is associated to @xmath258 by lemma [ lem : ext ]
. then the subgroups @xmath259 generate the asserted free product within @xmath260 by an application of the ping - pong lemma . finally , every @xmath261 is the unique automorphism of @xmath4 associated to @xmath262 by lemma [ lem : ext ] which can be realized as an element of @xmath263 by composing the inversions along the edges that occur in the unique reduced path from @xmath239 to @xmath264 .
lemma [ lem : ext ] also plays an important role in proving the following list of basic properties of the groups @xmath6 .
[ prop : uf_properties ] let @xmath1 .
then the following statements hold .
* @xmath6 is closed in @xmath265 .
* @xmath6 is locally permutation isomorphic to @xmath3 .
* @xmath6 is vertex - transitive .
* @xmath6 is edge - transitive if and only if the action @xmath266 is transitive .
* @xmath6 is discrete in @xmath244 if and only if the action @xmath266 is free .
proposition [ prop : uf_properties ] illustrates the principle that properties of @xmath6 correspond to properties of @xmath3 , which constitutes part of the beauty of the universal group construction . for ( i )
, suppose that @xmath267 .
then @xmath268 for some @xmath217 and hence the open neighbourhood @xmath269 of @xmath113 is also contained in the complement of @xmath6 in @xmath244
. for ( ii ) , let @xmath245 and @xmath270 .
further , let @xmath271 be the automorphism associated to @xmath272 .
then @xmath273 for every @xmath217 and hence @xmath274 realizes the permutation @xmath275 at the vertex @xmath239 . for part ( iii ) , let @xmath234 and let @xmath224 be the automorphism of @xmath4 associated to @xmath276 by lemma [ lem : ext ] .
then @xmath277 as @xmath278 for all @xmath217
. as to ( iv ) , suppose that @xmath3 is transitive .
given @xmath279 , choose @xmath280 such that @xmath281 by ( iii ) .
then pick @xmath282 such that @xmath283 , by ( ii ) and transitivity of @xmath3 , and set @xmath284 .
conversely , if @xmath6 is edge - transitive then @xmath285 acts transitively on @xmath216 for a given vertex @xmath217 and hence @xmath3 is transitive by ( ii ) . for part ( v ) , fix @xmath245 and suppose that the action @xmath266 is not free , say @xmath286 fixes @xmath252 . for every @xmath287 we define @xmath288 such that @xmath289 but @xmath290 as follows : set @xmath291 and let @xmath292 be an edge with @xmath293 , @xmath294 and @xmath295 .
then we may extend @xmath296 to @xmath297 as desired by setting @xmath298 and @xmath299 for all @xmath300 .
now , inductively extend @xmath296 to @xmath4 such that @xmath301 whenever @xmath302 maps to @xmath24 under the projection onto @xmath27 and @xmath303 otherwise .
conversely , assume that @xmath6 is non - discrete .
then there are @xmath288 and @xmath287 such that @xmath289 but @xmath290 .
hence there is @xmath304 such that @xmath305 is non - trivial and fixes a point .
for the sake of clarity we extract the following statement from proposition [ prop : uf_properties ] .
let @xmath1 .
then @xmath6 is compactly generated , totally disconnected , locally compact hausdorff .
it is discrete if and only if the action @xmath266 is free .
the group @xmath6 is totally disconnected , locally compact hausdorff as a closed subgroup of @xmath244 .
furthermore , @xmath6 is generated by the compact set @xmath306 where @xmath245 is a fixed vertex and @xmath307 is the edge - inversion of example [ ex : uf ] .
this follows from vertex - transitivity of @xmath308 : for @xmath288 pick @xmath309 such that @xmath310 .
then @xmath311 , hence the assertion .
since @xmath6 is vertex transitive it can not stabilize any subtree . since it is transitive on the ( directed ) edges of any given color it is easy to see that it can not fix any end of the tree .
thus @xmath6 is geometrically dense .
we claim that the group satisfies property @xmath171 . for every @xmath312 we have @xmath313 where @xmath314 is given by @xmath296 on @xmath315 and the identity elsewhere , and similarly for @xmath316 .
then for @xmath317 we have @xmath318 and @xmath319 otherwise .
this argument carries over to arbitrary finite or infinite paths in @xmath4 .
@xmath320(-0.5,0 ) \psline(-0.5,0)(0.5,0 ) \uput[d](0,0){$e$ } \psline(-0.5,0)(-1,0.5 ) \psline(-0.5,0)(-1,-0.5 ) \psdots[](0.5,0 ) \psdots[](-1,0.5 ) \psdots[](-1,-0.5 ) \rput(-1.5,0){$\cdots$ } \rput(-2.5,0){$t_{x}$ } \psline(0.5,0)(1,-0.5 ) \psline(0.5,0)(1,0.5 ) \psdots[](1,-0.5 ) \psdots[](1,0.5 ) \rput(1.5,0){$\cdots$ } \rput(2.5,0){$t_{y}$ } \psarc[linestyle = dashed](-2.5,0){2}{320}{40 } \psarc[linestyle = dashed](2.5,0){2}{140}{220 } \endpspicture\ ] ] as a consequence , the subgroup @xmath321 of @xmath6 is simple in many cases .
more precisely , we have the following .
let @xmath1 .
then @xmath322 is either trivial or simple . if @xmath3 is transitive and generated by point stabilizers we have @xmath323 and therefore @xmath324=2 $ ] .
the simplicity assertion follows from theorem [ tits - k ] . if @xmath3 is transitive and generated by edge stabilizers then @xmath322 acts transitively on geometric edges
this implies the non - trivial inclusion in the equality @xmath323 : namely , let @xmath325 and fix @xmath292 .
choose @xmath326 such that @xmath327 .
then @xmath328 and hence @xmath329 .
since @xmath330=2 $ ] , the non - trivial coset being given by an edge - inversion , this implies @xmath324\le 2 $ ] and equality follows from , say , the existence of edge - inversions in @xmath6 . the groups @xmath6 are universal in the following sense .
[ prop : universality ] let @xmath331 be vertex - transitive and locally permutation isomorphic to @xmath1 .
if @xmath3 is transitive then there is a legal labelling @xmath231 of @xmath4 such that @xmath332 .
fix @xmath245 .
since @xmath152 is locally permutation isomorphic to @xmath3 , there is a bijection @xmath333 such that @xmath334 . we now inductively define a legal labelling @xmath218 such that @xmath335 .
set @xmath336 and suppose inductively that @xmath231 is defined on @xmath337 . to extend @xmath231 to @xmath338 , let @xmath304 and let @xmath339 be the unique edge with @xmath340 and @xmath341 .
since @xmath152 is vertex - transitive and locally permutation isomorphic to the transitive group @xmath3 , there is an element @xmath342 which inverts the edge @xmath343 .
we may thus legally extend @xmath231 to @xmath216 by setting @xmath344 . to check the inclusion @xmath335 , let @xmath217 and @xmath345 .
if @xmath346 and @xmath347 denote the unique reduced paths from @xmath239 to @xmath24 and @xmath348 , then @xmath349 and we have @xmath350 by lemma [ lem : cocycle_identity ] .
in this section , we exhibit a point stabilizer in @xmath6 as a profinite group in terms of @xmath3 for transitive @xmath1 . to this end , let @xmath245 , @xmath351 , @xmath352 and set @xmath353 .
we fix bijections @xmath354 as follows : given that @xmath3 is transitive we may for every @xmath252 fix an element @xmath355 with @xmath356 .
define inductively * @xmath357 , and * @xmath358 for @xmath287 , where @xmath359 is the canonical projection .
we now capture the action of @xmath360 on @xmath361 by inductively defining @xmath362 as follows : let @xmath363 , set * @xmath364 , and define * @xmath365 to be the wreath product for the action of @xmath366 on @xmath367 .
further , let @xmath368 denote the canonical projection .
the bijection @xmath369 induces the surjective homomorphism @xmath370 with kernel @xmath371 and one readily checks that the map @xmath372(0.2,0.15)(-0.2,0.15)\psline[](-0.2,0.15)(-0.125,0.075)\endpspicture}}{\lim}\ f(n)=\left\{\left.(f_{n})_{n=1}^{\infty}\in\prod_{n=1}^{\infty}f(n)\right|\forall n\in\mathbb{n}:\pi_{n}f_{n}=f_{n-1}\right\}\ ] ] is an isomorphism of topological groups : abbreviate @xmath373(0.2,0.15)(-0.2,0.15)\psline[](-0.2,0.15)(-0.125,0.075)\endpspicture}}{\lim}\ f(n)}$ ] .
clearly , @xmath374 is a bijective homomorphism . to prove that it is a homeomorphism , note that @xmath360 is compact and @xmath375(0.2,0.15)(-0.2,0.15)\psline[](-0.2,0.15)(-0.125,0.075)\endpspicture}}{\lim}\ f(n)}$ ] is hausdorff ; therefore @xmath374 is closed and it suffices to show continuity . let @xmath376 be a basic open neighbourhood of @xmath377(0.2,0.15)(-0.2,0.15)\psline[](-0.2,0.15)(-0.125,0.075)\endpspicture}}{\lim}\ f(n)}$ ] . then there is @xmath378 such that @xmath379 for all @xmath380 and hence for every @xmath381 the open neighbourhood @xmath382 of @xmath113 is contained in @xmath383 .
the purpose of this section is to find new examples of t.d.l.c .
groups which are ( abstractly ) simple , compactly generated and non - discrete .
the motivation for this comes from classification results of locally compact groups by caprace and monod @xcite which yield cases in which compactly generated t.d.l.c .
groups decompose into ( topologically ) simple compactly generated non - discrete pieces .
we are still at the stage of collecting examples of such simple groups , aiming in the long term for some sort of classification .
the examples collected so far can be classified into the following types : * simple lie groups * simple algebraic groups over local fields * complete kac
moody groups over finite fields * automorphism groups of trees , some cat(0 ) cube - complexes , and right - angled buildings * variations on the above ( e.g. almost automorphisms of trees ) . to obtain these new examples of compactly generated , simple , non - discrete t.d.l.c . groups , we shall introduce some `` @xmath0-thickened '' ( for @xmath384 ) variations of the universal group construction : the _ @xmath0-closures _ of a given @xmath43 .
these prescribe the action on all balls of radius @xmath0 by elements of @xmath10 .
we will then see that the @xmath0-closure of a group of tree automorphisms satisfies property @xmath168 and use theorem [ tits - k ] to obtain abstractly simple t.d.l.c .
groups which are compactly generated .
the last step will be ensuring that these are non - discrete and different . in this setting , @xmath10 is the analogue of @xmath3 in the definition of @xmath6 , providing a list of `` allowed '' actions .
however , here we do not require that the tree be regular .
note that a given @xmath387 need not agree with the same element of @xmath10 on every ball ; the point is that for each ball there is _ some _ element of @xmath10 agreeing with @xmath144 , and they may all be different for each ball . the @xmath0-closure of @xmath10 has the following basic properties , which justify the term `` closure '' . * to see that @xmath388 is indeed a subgroup of @xmath42 , notice that if @xmath394 and @xmath25 then there exist @xmath395 such that @xmath396 and @xmath397 , so @xmath398 and @xmath399 .
also , there exists some @xmath400 such that @xmath401 , so @xmath402 and @xmath403 .
+ for the closure part , note that for each @xmath404 there is some vertex @xmath405 such that no element of @xmath10 agrees with @xmath275 on @xmath406 . thus @xmath407 is a union of basic open sets .
* the group @xmath10 agrees with itself on balls of all radii so @xmath408 for all @xmath231 and if @xmath391 then @xmath409 certainly agrees with @xmath10 on balls of smaller radius @xmath0 , so @xmath410 . * since @xmath411 for all @xmath0 and @xmath388 is closed , the closure of @xmath10 must be contained in every @xmath388 and therefore in the intersection of all of them . for the other direction ,
we show that every element @xmath412 is a point of closure of @xmath10 .
fix a vertex @xmath24 and consider @xmath413 ; then , since @xmath414 , there is some @xmath46 such that @xmath415 .
let @xmath417 be any finite or ( bi-)infinite path and suppose that @xmath418 . to see that @xmath419 , pick a vertex @xmath420 , which must be in @xmath421 for some @xmath422 . by definition ,
@xmath423 is the restriction to @xmath421 of some @xmath424 .
thus , if @xmath425 is entirely contained in @xmath421 then @xmath426 for some @xmath46 , since @xmath427 . and if there is some part of @xmath425 outside @xmath421 then both @xmath428 and @xmath429 act trivially on it . in either case , there is some @xmath46 such that @xmath430 and @xmath419 .
child [ label = below : ] foreach in 1,2,m node[draw , inner sep=1pt ] @xmath432 child[level distance=22 mm ] node edge from parent[snake = expanding waves , segment length=2mm , segment angle=60 , dashed , draw]node[fill = black!20]@xmath433 ; we know from [ a : closure ] that @xmath434 , so it suffices to show that @xmath435 for all @xmath436 . to illustrate the proof , we only show the case @xmath437 ( see ( * ? ? ?
* theorem 5.4 ) for the full proof ) .
let @xmath438 .
for each vertex @xmath420 there is some @xmath46 such that @xmath439 ; thus @xmath440 .
suppose that @xmath441 are the neighbours of @xmath420 .
since @xmath442 , for each @xmath211 there exist @xmath443 such that @xmath444 .
so @xmath445 fixes the edge @xmath446 .
because @xmath10 satisfies property @xmath172 , there exist unique @xmath447 such that @xmath448 , where @xmath449 only acts non - trivially on @xmath450 ( and @xmath451 fixes @xmath450 , see figure [ a : pk_stableclosures ] )
. then the product @xmath452 fixes all neighbours of @xmath420 and @xmath420 itself ; that is , it fixes @xmath453 . furthermore ,
@xmath451 fixes @xmath450 , so @xmath454 and hence @xmath455 for each @xmath211 . thus @xmath456 and @xmath457 . since @xmath458
we conclude that @xmath459 , as required . if @xmath11 does not satisfy property @xmath168 for any @xmath0 , then @xmath460 ( by proposition [ a : prop ] ) .
hence @xmath461 for all @xmath0 and therefore the sequence @xmath462 never stabilizes ; in particular there are infinitely many distinct @xmath0-closures @xmath388 of @xmath10 . for the converse
, we have that @xmath462 never stabilizes , therefore there is no @xmath0 such that @xmath463 and so @xmath10 does not satisfy @xmath168 for any @xmath0 .
* proposition 3.3.1 ) let @xmath464 be a finite @xmath465-transitive permutation group on the set @xmath466 and @xmath467 the stabilizer of a point under this action .
assume that @xmath467 is non - abelian and simple .
let @xmath8 be a @xmath5-regular tree and @xmath135 a vertex - transitive subgroup .
if @xmath107 is any vertex we have a map @xmath468 given by the action of @xmath469 on @xmath470 .
assume that @xmath471 .
write @xmath472 and consider the map @xmath473 given by the action of @xmath124 on @xmath474 .
we omit the proof of this theorem , but we quote it to emphasize that in some cases there are _ local conditions _ on a group @xmath10 which already imply the stabilization of its @xmath0-closures . indeed ,
if the action of @xmath10 on every 1-ball is contained in @xmath3 , then the case @xmath480 in the above theorem yields that @xmath481 .
let @xmath0 be a local field with integer ring @xmath482 , maximal ideal @xmath483 and residue field @xmath484 .
for the action of the group @xmath485 on its bruhat tits tree , the local action is given by the group @xmath486 and its action on the projective line @xmath487 .
below we discuss the fact that in this case the sequence of @xmath0-closures @xmath488 never stabilizes and hence gives rise to an infinite sequence of simple groups containing @xmath10 .
the theorem of burger
mozes above implies in particular that this kind of behaviour would never be possible when the `` local group '' @xmath3 is @xmath489 , for example .
\(i ) the group @xmath490 is generated by fixators of @xmath0-edges ( in particular , fixators of finite sets of vertices ) , which are basic open sets .
hence it is open .
+ ( ii ) this follows from the facts that all subgroups of discrete groups are discrete and that all open subgroups of a non - discrete group are non - discrete .
+ ( iii ) let @xmath158 be some path in @xmath8 .
since @xmath388 satisfies @xmath168 we have that @xmath491 and @xmath492 suppose that @xmath495 and let @xmath497 be a generator ( so that it fixes some edge @xmath498 ) .
now , @xmath499 satisfies @xmath168 and therefore @xmath500 where @xmath501 fixes @xmath502 pointwise and @xmath503 fixes @xmath504 pointwise .
in particular , there exist edges @xmath505 and @xmath506 such that @xmath507 fixes @xmath508 and @xmath509 fixes @xmath510 .
hence @xmath511 are generators of @xmath512 and @xmath513 , as required .
conversely , if @xmath493 then for any @xmath514 and any vertex @xmath420 there is some @xmath46 such that @xmath515 fixes @xmath425 .
in particular , @xmath515 fixes @xmath516 where @xmath517 is some edge starting at @xmath420 .
thus @xmath518 , that is @xmath519 .
since @xmath520 we conclude that @xmath521 .
thus , in order to construct infinitely many distinct t.d.l.c .
simple non - discrete subgroups of aut@xmath522 it suffices to find examples with infinitely many distinct @xmath0-closures . by corollary [ stable_k_closures
] , this amounts to finding examples which do not satisfy property @xmath168 for any @xmath0 .
recall that @xmath8 , the bass serre tree of @xmath526 , has vertices the left ( say ) cosets of @xmath527 and ( directed ) edges labelled by @xmath528 from @xmath529 to @xmath530 if and only if there is some @xmath211 such that @xmath531 .
@xmath532 acts on @xmath8 by left multiplication and , in the non - solvable case ( when neither @xmath533 nor @xmath28 equals 1 ) , this action is geometrically dense .
we claim that when @xmath534 are coprime @xmath535 does not satisfy @xmath168 for any @xmath0 . to see
this ( figure [ a : bass - serre ] may be helpful ) , consider the edge @xmath536 .
we will show that @xmath537 while @xmath538 ( it contains , for instance , @xmath539 ) .
let @xmath540 , then @xmath24 must also fix @xmath527 as all other neighbours of @xmath541 are fixed by @xmath24 .
thus @xmath24 must be of the form @xmath542 where @xmath543 is a multiple of @xmath28 ; say @xmath544 for some @xmath545 with @xmath546 and @xmath227 not divisible by @xmath28 .
note that @xmath547 contains vertices of the form @xmath548 for all @xmath549 .
pick some @xmath550 .
if @xmath544 fixes @xmath551 then , since @xmath552 , we have @xmath553 . for the last equality to hold
, @xmath554 must be a multiple of @xmath28 , which it can not be by the assumption that @xmath534 are coprime and the choice of @xmath227 , unless @xmath555 .
thus we must have @xmath556 .
a similar argument yields that @xmath557 .
( root ) @xmath558 child node[circle , draw ] ( t ) @xmath559 edge from parent[-latex ] child node[circle , draw ] ( at ) @xmath560 edge from parent[-latex ] child node[circle , draw ] ( t-1 ) @xmath561 edge from parent[latex- ] child node[circle , draw ] ( t-1at ) @xmath562 edge from parent[-latex ] child node[circle , draw ] ( t-2 ) @xmath563 child node[circle , draw ] ( t-1at-1 ) @xmath564 child node[circle , draw ] ( t-1at-1at-1 ) @xmath565 child node[circle , draw ] ( t-1a2 t ) @xmath566 edge from parent[latex- ] child node[circle , draw ] ( at-1 ) @xmath567 edge from parent[latex- ] child node[circle , draw ] ( a2t-1 ) @xmath568 edge from parent[latex- ] ; the group @xmath569 acting on its bruhat tits tree ( which is isomorphic to @xmath570 ) also does not satisfy @xmath168 for any @xmath0 . indeed it is well known that the action of @xmath10 on @xmath38 is isomorphic to the action of the same group on the projective line @xmath571 by fractional linear transformations .
in particular the stabilizer of three boundary points is trivial .
this means that an element of @xmath10 is completely determined by its action on three distinct points of @xmath38 ( these elements should be thought of as @xmath572-adic mbius transformations ) .
now let @xmath573 be an infinite or finite geodesic . and assume that @xmath574 with @xmath575 .
then each such @xmath423 is defined on @xmath576 which contains many boundary points .
hence each @xmath423 admits a unique extension to the whole tree .
this is a strong obstruction to satisfying property @xmath168 , for any @xmath0 .
we note that this method finds infinitely many t.d.l.c .
simple non - discrete groups which are pairwise distinct as subgroups of aut@xmath522
. it would be desirable to know whether these subgroups are pairwise non - isomorphic .
this is stated as work in progress in @xcite . using different methods , s. smith ( @xcite )
has found uncountably many t.d.l.c .
simple non - discrete groups which are pairwise non - isomorphic .
this is discussed by c. reid and g. willis in their chapter . | this article is an expanded version of the talks given by the authors at the arbeitsgemeinschaft `` totally disconnected groups '' , held at oberwolfach in october 2014 .
we recall the basic theory of automorphisms of trees and tits simplicity theorem , and present two constructions of tree groups via local actions with their basic properties : the universal group associated to a finite permutation group by m. burger and s. mozes , and the @xmath0-closures of a given group by c. banks , m. elder and g. willis . |
soon after the discovery of pulsars , it was shown that the radio emission from pulsars is highly polarized @xcite . the mean pulse profile and polarization properties of a pulsar
are important for understanding the pulse emission mechanism , the beaming of pulsar radiation and the geometry of the system .
mean pulse profiles often have a double or triple form , leading to a description of the emission in terms of `` core '' and `` conal '' beams @xcite .
@xcite suggested that the core and conal components had different emission mechanisms whereas @xcite argued that there was a gradual change in emission characteristics from the core region to the outer edge of the emission beam rather than distinct emission processes . to explain more complex pulse profiles , multiple emission cones were proposed and discussed by several authors @xcite .
an alternative model suggests that the emission beam of a pulsar is filled with randomly distributed emission patches @xcite .
the patchy model may explain more complicated multi - component and asymmetric pulse profiles .
polarization properties of radio sources are normally described in terms of the stokes parameters : @xmath0 ( total intensity ) , @xmath1 and @xmath2 ( linear polarization ) and @xmath3 ( circular polarization ) .
many pulsars show a systematic variation of the position angle ( pa , @xmath4 ) of the linearly polarized emission , @xmath5 across the pulse profile . in many pulsars
the observed pa variations can be approximately described by the rotating vector model " ( rvm ) @xcite .
this model originates from the idea that the radiation is polarized in the plane of curvature of field lines emanating from a magnetic pole on the star ; not surprisingly , curvature radiation has this property @xcite . for a simple dipole field , the observed pa variation
is then determined by the projected direction of the magnetic axis as the star rotates .
the rapid swings often observed near the profile mid - point imply that magnetic axis is nearly aligned with the observer s line of sight at that profile phase .
circular polarization ( stokes @xmath3 ) is usually relatively weaker than linear polarization .
it is most often associated with the central or core component of the profile , often with a sense reversal near the profile mid - point @xcite .
observed pa swings are not always smooth and continuous .
discontinuities of approximately 90 are often observed ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and these are interpreted as resulting from overlapping emission from orthogonally polarized emission modes ( e.g. , * ? ? ?
such orthogonal modes may be generated as the wave propagates through the pulsar magnetosphere ( e.g. , * ? ? ?
* ) . in many respects , millisecond pulsars ( msps )
have very different properties to those of `` normal '' pulsars . besides their obviously shorter periods , they have much smaller period derivatives and hence larger characteristic ages and weaker implied dipole magnetic fields .
the pulse profile for millisecond pulsars generally has a much larger duty cycle , i.e. , the emission covers a much larger fraction of the pulse period , and a more complicated pulse profile compared to normal pulsars ( see , e.g. , * ? ? ?
* ; * ? ? ?
this might be expected within the context of the magnetic - pole model since the pulsar magnetosphere , bounded by the light cylinder of radius @xmath6 is so much smaller .
consequently , the polar caps defined by the open field lines ( those that penetrate the light cylinder ) are larger and the opening angle of these open field lines at a given radial distance @xmath7 and fractional polar - cap radius @xmath8 is also larger : @xmath9 where @xmath10 is the field - line tangent angle with respect to the magnetic axis ( see e.g. , @xcite ) . despite these differences ,
the polarization characteristics of msps are remarkably similar to those of normal pulsars they have similar high degrees of linear polarization , orthogonal - mode pa jumps are observed and circular polarization with sense changes near the profile centre is seen in some msps ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
although the emission characteristics are similar , indicating that the emission mechanism is basically the same for normal pulsars and msps , the pa variations in msps are often more complex and do not fit the rvm well .
faraday rotation occuring in the interstellar medium is relatively easy to observe in pulsars because of the typically high degree of linear polarization .
faraday rotation is quantified by the rotation measure ( rm ) , defined by @xmath11 where @xmath12 is the radio wavelength corresponding to radio frequency @xmath13 .
faraday rotation in the interstellar medium is given by : @xmath14 where @xmath15 is the vector magnetic field in microgauss and the integral is over the path to the pulsar , and so observations of faraday rotation can be used to investigate the galactic magnetic field ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
pulsars have the unique advantage that the dispersion measure @xmath16 is also known .
we can therefore obtain a measure of the mean line - of - sight component of the interstellar magnetic field weighted by the local electron density along the path : @xmath17 where rm and dm are in their usual units .
in this paper we report on the mean pulse profiles and polarization properties of the 20 msps which are being observed as part of the parkes pulsar timing array ( ppta ) project @xcite .
details of the observing system and the observations are given in
[ sec : obs ] and polarization profiles for the 20 pulsars are presented in [ sec : poln ] .
rm results are presented in [ sec : rm ] and the implications of the results are discussed in [ sec : discn ] .
as part of the ppta project , frequent observations of 20 msps are made using the parkes 64-m radio telescope .
regular observations at approximately three - weekly intervals commenced in mid-2004 .
since that time , back - end signal processing systems have been upgraded several times .
the main objective of these observations is to detect gravitational waves through high - precision pulsar timing .
however , there are many secondary objectives , including studies of the pulse emission properties for pulsars in the observed sample .
all ppta data are recorded in full - polarization mode , not only enabling full profile calibration , but also making possible detailed studies of the pulse emission properties .
this paper reports on polarization results from observations made with the centre beam of the parkes 20 cm multibeam receiver @xcite with the second - generation parkes digital filterbank system pdfb2 between 2007 june and 2009 november .
the fact that many observations are available for all 20 pulsars means that pulse profiles with very high signal / noise ratios can be formed , leading to new information about the profile shapes and their polarization .
feeds in the parkes 20 cm multibeam receiver have orthogonal linearly polarized probes with a calibration probe at 45 to the two signal probes through which a linearly polarized broad - band and pulsed calibration signal can be injected .
the system equivalent flux density on cold sky is approximately 30 jy . in common with the other parkes digital filterbank systems , pdfb2 digitizes band - limited signals from each of the two orthogonal probes at the nyquist rate with 8-bit sampling .
it uses field - programmable gate array ( fpga ) processors to implement a polyphase filter to provide frequency resolution and a synchronous averager to form mean pulse profiles . for the observations reported here the total bandwidth
was 256 mhz centred at 1369 mhz with 1024 channels across the band for most of the 20 pulsars .
for all except psrs j1857 + 0943 , j1939 + 2134 and j2124@xmath183358 , each observation was of 64 minutes duration ; for the exceptions , the observation times were 32 minutes .
all data were recorded using the psrfits data format @xcite with 1-minute sub - integrations and the full spectral resolution . to provide a calibration of the gain and phase of the receiver system , the pulsed calibration signal was recorded for 2 minutes prior to each pulsar observation .
signal amplitudes were placed on a flux density scale using observations of hydra a , assumed to have a flux density of 43.1 jy at 1400 mhz and a spectral index of @xmath180.91 .
the 20 cm multibeam feeds have limited isolation between the orthogonal polarizations .
this cross - coupling was measured using series of observations of psr j0437@xmath184715 covering a wide range of hour angles , allowing correction for the effects of this coupling @xcite .
off - line data processing made use of the psrchive pulsar data analysis system @xcite .
first , band edges ( 5% on each side ) and spurious data resulting from narrow - band and impulsive radio - frequency interference were effectively excised from the raw data files .
typically 1 2% of the spectrum and the time - domain data were excised .
the data were then averaged in time to give eight sub - integrations for each observation .
next , the data for each observation were corrected for variations in instrumental gain and phase across the spectrum and for the effects of cross - coupling in the feed . finally , baselines for each of the stokes - parameter profiles were set to zero mean and the data placed on a flux density scale .
the baseline region was computed from the stokes @xmath0 profile and applied to all four stokes - parameter profiles . following this calibration ,
stokes parameters were in accordance with the astronomical conventions described by @xcite .
specifically , pas are absolute and are measured from celestial north toward east , i.e. , counterclockwise on the sky , and stokes @xmath3 is defined as @xmath19 , using the ieee definition for sense of circular polarization . in order to form final mean polarization profiles , it is necessary to sum across the frequency channels taking into account the faraday rotation across the band .
sufficiently precise rm measurements were not previously available for most of the ppta sample .
we therefore determined rm values for each pulsar from the pdfb2 data sets .
for each individual observation , the upper and lower halves of the total band were separately summed using the nominal rm value .
an improved value of the rm was then calculated by taking a weighted mean of position angle differences for bins where the uncertainty in position angle was less than 10 .
this process was iterated until convergence was obtained .
the contribution of the earth s ionosphere to the total rm was estimated and then subtracted so that the measured rm just represents the interstellar contribution .
ionospheric rms were computed using a prediction program farrot , developed at the dominion radio astrophysical observatory , penticton , which uses the observed solar 10.7-cm radio flux as a basic input .
as the observations were made near a minimum of solar activity , ionospheric rms are relatively small , typically between @xmath20 and @xmath21 rad m@xmath22 .
the final interstellar rm for each pulsar was determined by taking a weighted mean of the individual interstellar rm measurements .
in addition , the observed stokes parameters for each observation were adjusted ( effectively a pa shift ) using the ionospheric rm so that they represent the polarization state at the top of the ionosphere .
it is also necessary to have a precise timing model for each pulsar before adding the data in time to form a final mean profile .
pulse times of arrival were obtained for each observation using an analytic template based on an existing high signal / noise ( s / n ) ratio pulse profile .
the tempo2 pulsar timing package @xcite was then used to fit pulsar spin and astrometric and binary parameters as necessary to give `` white '' timing residuals for the pdfb2 data set for each pulsar .
finally , the separate observations were summed using this timing model to determine relative phases to form the final stokes parameter profiles . to give the best possible s / n ratio in the final profile , the individual observation profiles were weighted by their ( s / n)@xmath23 when forming the sum profile .
the polarization parameters @xmath24 , @xmath25 and pa were then computed from the stokes parameters .
the noise bias in @xmath26 was corrected using the relations given by @xcite and the similar bias in @xmath25 was corrected using @xmath27 when @xmath28 or zero otherwise , where @xmath29 is the rms profile baseline noise .
table [ tb : obs ] gives the basic pulsar and observational parameters for the 20 ppta pulsars which were observed .
after the pulsar name , pulse period and dm , we give the number of frequency channels across the 256-mhz band and the number of bins across the pulse period . the dispersion smearing across each frequency channel , given by @xmath30 where @xmath31 is the channel width in mhz
, @xmath32 is the band centre frequency in mhz , and the dm is in units of @xmath33pc is given in units of profile bins , the number of observations summed , and the total observation time are given in the remaining three channels .
cccccccc & @xmath34 & dm & nr of & nr of & dm smear & nr of & integ .
time + & ( ms ) & ( @xmath33 pc ) & channels & bins & ( bins ) & obs . &
( h ) + j0437@xmath184715 & 5.757 & 2.64 & 1024 & 1024 & 0.4 & 92 & 96.3 + j0613@xmath180200 & 3.062 & 38.78 & 1024 & 512 & 5.2 & 41 & 43.3 + j0711@xmath186830 & 5.491 & 18.41 & 1024 & 512 & 1.4 & 28 & 29.1 + j1022 + 1001 & 16.453 & 10.25 & 1024 & 2048 & 1.0 & 34 & 34.3 + j1024@xmath180719 & 5.162 & 6.49 & 1024 & 1024 & 1.0 & 29 & 30.0 + + j1045@xmath184509 & 7.474 & 58.17 & 2048 & 512 & 1.6 & 33 & 34.1 + j1600@xmath183053 & 3.598 & 52.33 & 1024 & 512 & 6.0 & 28 & 29.7 + j1603@xmath187202 & 14.842 & 38.05 & 1024 & 1024 & 2.1 & 23 & 23.7 + j1643@xmath181224 & 4.622 & 62.41 & 1024 & 512 & 2.8 & 34 & 36.1 + j1713 + 0747 & 4.570 & 15.99 & 1024 & 1024 & 2.9 & 40 & 40.7 + + j1730@xmath182304 & 8.123 & 9.62 & 1024 & 1024 & 1.0 & 23 & 23.7 + j1732@xmath185049 & 5.313 & 56.82 & 2048 & 512 & 2.2 & 24 & 24.7 + j1744@xmath181134 & 4.075 & 3.14 & 512 & 1024 & 1.3 & 33 & 34.4 + j1824@xmath182452 & 3.054 & 120.50 & 2048 & 256 & 4.1 & 27 & 28.0 + j1857 + 0943 & 5.362 & 13.30 & 1024 & 1024 & 2.1 & 26 & 13.4 + + j1909@xmath183744 & 2.947 & 10.39 & 1024 & 512 & 1.5 & 67 & 69.6 + j1939 + 2134 & 1.558 & 71.04 & 1024 & 256 & 9.4 & 26 & 13.8 + j2124@xmath183358 & 4.931 & 4.60 & 1024 & 1024 & 0.8 & 25 & 13.1 + j2129@xmath185721 & 3.726 & 31.85 & 1024 & 512 & 3.5 & 25 & 25.5 + j2145@xmath180750 & 16.052 & 9.00 & 1024 & 2048 & 0.9 & 29 & 29.6 +
polarization profiles at 1369 mhz for the 20 ppta msps are presented and discussed in this section .
table [ tb : poln ] gives a summary of the results .
columns 2 and 3 give the mean flux density @xmath35 averaged over all observations and its uncertainty .
the next column gives the rms fluctuation in individual - observation flux densities .
for low dm pulsars ( dm @xmath36 @xmath33 pc ) the rms fluctuation is comparable to or even greater than the mean value .
this entirely results from intensity modulations due to diffractive interstellar scintillation .
pulse widths at 50% of the peak flux density ( w50 ) and 10% of the peak ( w10 ) are given next , both in degrees of longitude , where @xmath37 is equivalent to the pulse period or 1.0 in pulse phase , and in milliseconds .
the next column gives the overall pulse width , measured from the first point to the last point where the pulse intensity significantly exceeds the baseline noise ( more than three times the baseline rms noise in several adjacent bins ) .
the next column gives the minimum number of identifiable pulse components in the pulse profile .
a component is identified by a peak or , for overlapping components a significant inflection , in the pulse profile .
the next three columns give the fractional linear polarization @xmath38 , the fractional net circular polarization @xmath39 and the fractional absolute circular polarization @xmath40 averaged over all observations , where the @xmath41 means are taken across the pulse profile .
cccccccccccccc & @xmath42 & @xmath43 & @xmath44 & & & & overall width & nr of & @xmath45 & @xmath46 & @xmath47 + & ( mjy ) & ( mjy ) & ( mjy ) & ( deg ) & ( ms ) & & ( deg ) & ( ms ) & ( deg ) & comp . &
( % ) & ( % ) & ( % ) + j0437@xmath184715 & 149.3 & 4.0 & 38.0 & 9&0.14 & & 64 & 1.02 & 306 & 12 & 24 & @xmath185 & 11 + j0613@xmath180200 & 2.3 & 0.1 & 0.6 & 55 & 0.92 & & 108 & 8.18 & 158 & 7 & 15 & 2 & 4 + j0711@xmath186830 & 1.4 & 0.3 & 2.0 & 124&1.90 & & 169 & 2.57 & 277 & 10 & 4 & @xmath183 & 4 + j1022 + 1001 & 1.5 & 0.3 & 2.0 & 21 & 0.98 & & 43 & 1.96 & 68 & 6 & 51 & @xmath1811 & 14 + j1024@xmath180719 & 1.5 & 0.3 & 1.7 & 35 & 0.50 & & 106 & 1.51 & 158 & 9 & 55 & 2 & 7 + + j1045@xmath184509 & 2.2 & 0.1 & 0.5 & 37 & 0.76 & & 70 & 1.45&248 & 9 & 20 & 13 & 14 + j1600@xmath183053 & 2.4 & 0.1 & 0.4 & 9 & 0.09 & & 41 & 0.42 & 68 & 3 & 27 & 2 & 3 + j1603@xmath187202 & 4.2 & 0.7 & 3.4 & 29 & 1.21 & & 41 & 1.71 & 238 & 7 & 17 & 27 & 28 + j1643@xmath181224 & 5.0 & 0.7 & 4.2 & 25 & 0.32 & & 72 & 0.93 & 241 & 5 & 11 & @xmath181 & 9 + j1713 + 0747 & 7.4 & 1.1 & 7.4 & 9 & 0.11 & & 30 & 0.39 & 104 & 5 & 28 & 0 & 3 + + j1730@xmath182304 & 3.9 & 0.3 & 1.5 & 43&0.98 & & 77&1.74 & 232 & 9 & 27 & @xmath1820 & 20 + j1732@xmath185049 & 1.3 & 0.1 & 0.5 & 20 & 0.29 & & 119 & 1.76 & 155 & 4 & 7 & 0 & 2 + j1744@xmath181134 & 3.3 & 0.6 & 3.5 & 12 & 0.14 & & 22 & 0.25 & 209 & 8 & 87 & 1 & 3 + j1824@xmath182452 & 1.6 & 0.1 & 0.6 & 115 & 0.98 & & 189 & 1.60 & 281 & 6 & 58 & 1 & 6 + j1857 + 0943 & 5.9 & 0.8 & 4.1 & 35 & 0.52 & & 203 & 3.03 & 238 & 6 & 4 & 2 & 2 + + j1909@xmath183744 & 2.6 & 0.2 & 2.0 & 5 & 0.04 & & 11 & 0.09 & 187 & 3 & 47 & 12 & 13 + j1939 + 2134 & 13.8 & 0.9 & 4.6 & 15 & 0.06 & & 198 & 0.86 & 302 & 6 & 27 & 1 & 2 + j2124@xmath183358 & 2.4 & 0.4 & 2.1 & 38 & 0.53 & & 275 & 3.76 & 331 & 12 & 18 & 0 & 3 + j2129@xmath185721 & 1.6 & 0.1 & 0.7 & 26 & 0.27 & & 60 & 0.62 & 144 & 5 & 41 & @xmath1823 & 26 + j2145@xmath180750 & 9.3 & 1.5 & 7.9 & 8 & 0.34 & & 94 & 4.17 & 187 & 7 & 10 & 6 & 6 + in the following subsections , we present polarization profiles and discuss each pulsar in turn .
mean pulse profile and polarization parameters for this very strong southern msp ( mean flux density about 145 mjy , table [ tb : poln ] ) at 1369 mhz are given in fig .
[ fig:0437 ] .
our results are in good agreement with and extend previously published results @xcite .
as is well known , the total intensity and polarization variations across the pulse profile are complex .
there are multiple overlapping components and the present results show that the pulse profile extends over at least 85% of the pulse period .
the notches in stokes @xmath0 and @xmath26 discussed by @xcite and @xcite are clearly visible in the central expanded plot .
the observed pa variations can not be described by the rvm , even if orthogonal mode jumps are taken into account .
there is a clear orthogonal mode transition in both linear and circular polarization very close to the main profile peak and a non - orthogonal pa transition around pulse phase @xmath48 @xcite .
we also observe a probable non - orthogonal transition near pulse phase 0.35 .
[ fig:0613 ] shows the polarization profiles for psr j0613@xmath180200 .
these are consistent with and extend the 1341 mhz results presented by @xcite .
@xcite observed very different profiles at 410 mhz and 610 mhz , indicating that there is significant profile morphology evolution between these lower frequencies and the 1369 mhz profile presented here .
in particular , the trailing peak is much stronger at low frequencies relative to the rest of the profile .
our results have higher s / n ratio than those previously published , and show that the pulse profile consists of multiple overlapping components ; a weak trailing component was not obvious in previous results .
significant linear polarization is seen across essentially the whole profile , showing that the pa variations are more complex than was evident from earlier work , with several regions of increasing and decreasing pa across the profile .
there is evidently a weak pulse component at the leading edge of the trailing component ( at phase 0.12 ) which has a distinctly different pa , discontinuous with the rest of the trailing component .
two clear peaks in circular polarization are seen , under the leading edges of the central broad pulse peak and the trailing peak .
there is a sense reversal in the circular polarization within the trailing component .
the mean pulse profile for this pulsar is basically a wide double with a connecting bridge of emission @xcite .
however , the results of @xcite showed that there were several overlapping components in the main peaks of the profile and furthermore that there was evidence for weak emission following the second main component , with significant emission over 75% of the pulse period .
these features are confirmed by the observations shown in fig .
[ fig:0711 ] . as well as the component preceding the first peak
, there is clear emission following the second peak , with a low double - peaked feature close to phase 0.3 in fig .
[ fig:0711 ] .
the polarization properties from our observations are very similar to those of @xcite with both linear and circular peaks centred under the main components .
however , we find evidence for an orthogonal mode transition just after the peak of the leading main component .
mean pulse profile and polarization parameters for this pulsar at 1369 mhz are given in fig .
[ fig:1022 ] .
the polarization profiles shown here are generally in good agreement with those presented by @xcite , @xcite and @xcite except that the fractional linear polarization in the trailing component in our observations is a little higher . given that the overall pa variation closely fits the rvm , it seems probable that the pa glitch near the centre of the profile results from a separate emission component .
similar central pa discontinuities are seen in several pulsars , e.g. psr b1913 + 16 @xcite and psr j1141@xmath186545 @xcite .
alternatively , it may result from a pair of closely spaced and smeared orthogonal - mode transitions .
our results , presented in fig
. [ fig:1024 ] , have a higher s / n ratio compared to earlier work ( e.g. , * ? ? ? * ) and show that the pulse profile has multiple overlapping features .
the leading part of the profile has extremely high fractional linear polarization whereas the trailing part of the profile is essentially unpolarized .
circular polarization is weak except near the centre of the profile where there is a sense reversal .
there is sub - structure in the circular polarization with at least three overlapping components in the positive @xmath3 part . in agreement with previous work
we find that there is little or no variation in pa across the profile .
relatively low - quality polarization profiles have been presented for this pulsar by @xcite and @xcite .
the data shown in fig . [ fig:1045 ]
are have higher s / n ratio and show a number of previously unrecognised profile features .
most notable is the leading emission around phase @xmath20 which is joined to the main pulse by a low - level bridge of emission .
this low - level emission extends the overall pulse width to more than 60% of the pulse period .
the polarization properties shown in fig . [ fig:1045 ]
are consistent with those given in the earlier papers but show more detail .
in particular , the pa variation is more complex with non - monotonic variations and a probable orthogonal transition near the leading edge of the main pulse .
there is a circular sense reversal between the first and second strong components of the main pulse .
mean pulse profile and polarization parameters for this pulsar at 1369 mhz are shown in fig .
[ fig:1600 ] .
our observations are in complete agreement with those of @xcite .
two orthogonal pa transitions are seen .
the second of these slightly trails the strongest profile peak and is coincident with a sense reversal in @xmath3 .
this situation is virtually identical to that seen in psr j0437@xmath184715 , with a fully orthogonal polarization transition very close to the main pulse peak .
mean pulse profile and polarization parameters presented in fig .
[ fig:1603 ] show that the profile for this pulsar has two main components , with the trailing one having very high circular polarization ( cf .
* ; * ? ? ?
table [ tb : poln ] shows that this pulsar has the highest fractional mean @xmath3 and @xmath25 of the ppta pulsars . the expanded central plot in fig .
[ fig:1603 ] shows that there is a broad low - level feature preceding the main pulse and a double - peaked pulse trailing the main pulse by about 0.4 in phase . with these additional pulse components ,
the pulse profile extends over about 70% of the period .
variations in polarization across the pulse are complex with three sense reversals in @xmath3 and an unusual pa variation which bears no relation to that expected for rvm .
the pulse polarization profiles given in fig .
[ fig:1643 ] have far more detail than previously published results @xcite .
they show that the pulse profile has a long leading ramp ( phase @xmath49 to @xmath50 in fig .
[ fig:1643 ] ) followed by a broad component leading into the main pulse which has at least three overlapping components .
there are two orthogonal polarization transitions near the pulse peak and these are associated with sense reversals in @xmath3 and dips in @xmath26 .
when account is taken of the two discontinuities , there is a decreasing pa through the stronger part of the profile , but with a significantly steeper pa swing in the trailing part .
the pa of the broad feature preceding the main pulse is not very well determined , but it appears discontinuous with the rest of the pa variations .
[ fig:1713 ] shows that this pulsar has a multi - component mean pulse profile extending over a little more than 0.25 in phase .
our results are consistent with , but improve on , those of @xcite . both the leading and trailing parts of the pulse profile have very high , almost complete , linear polarization .
similar to psr j1643@xmath181224 , there are two orthogonal pa transitions near the pulse peak .
the second one of these is clearly associated with a sense change in @xmath3 and the first may be also .
a third pa transition is observed preceding the trailing pulse component .
however , this is clearly not orthogonal , with @xmath51 .
mean pulse profile and polarization parameters for this pulsar at 1369 mhz are given in fig .
[ fig:1730 ] .
these show that the pulse profile is very complex , with multiple components in the main part , two weak leading components ( around phases @xmath52 and @xmath53 ) and a slightly stronger trailing component ( around phase 0.32 ) .
the weak trailing component was also observed by @xcite but the leading components have not been previously identified .
this pulsar is unusual in that it has relatively strong right - circular polarization through much of the main pulse ( cf .
the observed pa variation is very complex with a possible orthogonal transition in the leading part of the main pulse .
the rest of the profile shows complex pa variations with zones of both increasing and decreasing pa .
both the second precursor component and the postcursor are highly linearly polarized , probably with increasing pa across them .
the leading precursor is too weak to be sure about its polarization .
psr j1732@xmath185049 is a relatively weak pulsar .
results presented in fig .
[ fig:1732 ] are similar to those given by @xcite except that the fractional linear polarization is a little higher .
no significant circular polarization is observed across the profile .
the position angle swing is relatively flat with two pa jumps , both close to orthogonal . as shown in fig .
[ fig:1744 ] , the mean pulse profile of this pulsar has a sharp main pulse and a precursor preceding the main pulse by @xmath54 in phase .
the precursor was first detected by @xcite ; these authors also found evidence for a postcursor following the main pulse by about 0.2 in phase and amplitude about half of that of the precursor .
we do not see this postcursor component .
[ fig:1744 ] clearly shows that the precursor has multiple components .
this pulsar is remarkable in that the main pulse is almost 100% linearly polarized , especially on the leading edge .
however weak left - circular emission is observed in both the precursor and the main pulse .
there is a smooth decrease of pa through the main pulse .
this may be continuous with the pa of the precursor , indicating an outer line of sight , i.e. , that the observer s line of sight is equatorward of the magnetic axis . as fig .
[ fig:1824 ] shows , this pulsar , which is located in the globular cluster m28 @xcite , has a complex profile with multiple components covering more than 80% of the pulse period .
comparison with the results of @xcite shows that the leading strong component in fig .
[ fig:1824 ] has a very steep spectrum as it is by far the strongest component at 610 mhz .
our results are generally in agreement with those of @xcite but show more detail , including a weak component around phase @xmath55 . most of the strongest component in fig .
[ fig:1824 ] is 100% linearly polarized .
in fact , there is slight overpolarization of the ramp preceding this component .
this may result from the difficulty in accurately estimating baselines for this very wide profile or from the existence of polarized unpulsed emission from the pulsar .
the first strong component is also very highly polarized .
the pa variation across these components appears continuous , but the pa of the extreme leading and trailing components is not continuous with the central part , having the opposite slope . our results , shown in fig .
[ fig:1857 ] are in excellent agreement with those of @xcite showing a relatively weakly polarized main pulse and interpulse , both of which have multiple components . @xcite observed a weak steep - spectrum component just preceding the main pulse and we see this component as well ( at phase @xmath56 in fig .
[ fig:1857 ] ) .
the pa variation is complex across the main pulse and completely inconsistent with the rvm .
there is evidence for an orthogonal mode transition near the highest peak of the interpulse ( cf .
the mean pulse profile for this pulsar has a narrow main pulse and a weak feature preceding the main pulse by about 0.45 in phase .
our observations , shown in fig
. [ fig:1909 ] , are generally consistent with those of @xcite but show some additional features .
the `` interpulse '' feature is clearly seen to consist of two components and there is evidence for a weak shelf of emission immediately preceding the main pulse .
the main pulse is moderately polarized and has a pa variation which is almost consistent with the rvm and a very small impact parameter . however , the total pa swing is closer to @xmath57 rather than the expected @xmath58 .
this and the coincident sense reversal of circular polarization suggest that the rapid swing is actually a smeared orthogonal transition .
the smearing is not instrumental and must be intrinsic to the emitted radiation .
the mean pulse profile of psr j1939 + 2134 , the first - discovered msp @xcite , has a main pulse and a relatively strong interpulse separated by approximately @xmath59 or 0.48 in pulse phase . because of the high dm / p of this pulsar , our observations shown in fig .
[ fig:1939 ] are significantly affected by dm smearing ( see table [ tb : obs ] ) . as a result , we do not see the secondary maxima at the trailing edges of both pulses seen by @xcite , @xcite and @xcite all of whom used coherent dedispersion .
however , the high sensitivity of our observations reveals previously undetected pulsed emission preceding both the main pulse and the interpulse by about 0.2 in phase .
this emission has a peak flux density about 0.5% of that of the main pulse and appears to have multiple components . at such a low level , one
is concerned that otherwise undetected radio frequency interference ( rfi ) or instrumental problems may be responsible . however ,
when the data set is split in both frequency and time , the low - level components are visible in all of the partial data sets .
although low - level features have been detected in a number of pulsars in this study , in general they do not have the same relationship to the main pulse emission as is seen for psr j1939 + 2134 .
furthermore , for psr j1939 + 2134 , the emission preceding the interpulse has a different profile to that preceding the main pulse . for these reasons
, we believe that the low - level emission seen in fig .
[ fig:1939 ] is real and not either an instrumental artifact or due to rfi .
the polarization properties shown in fig . [ fig:1939 ]
are generally consistent with previously published polarimetry @xcite although we see more significant left - circular emission , especially in the main pulse .
all three of the earlier papers referred to above find evidence for an orthogonal mode transition near the leading edge of the main pulse .
however , our data suggest that the jump is closer to @xmath60 and therefore not orthogonal . on the other hand , there is some evidence for a reversal in the sense of circular polarization at the phase of the pa transition .
we also see a previously undetected orthogonal transition near the trailing edge of the interpulse . @xcite and @xcite find evidence for an orthogonal jump at low frequencies near the leading edge of the interpulse , but no evidence for jumps at frequencies around 1400 mhz . as fig .
[ fig:2124 ] shows , this pulsar has a extremely complicated multi - component mean pulse profile with emission covering nearly the entire pulse period .
the polarization profiles presented here are in good agreement with those presented by @xcite and @xcite for frequencies around 1400 mhz .
the fractional linear polarization changes greatly across the profile with the emission near phase @xmath55 being essentially 100% linearly polarized , whereas , on average , the profile is about 18% polarized ( table [ tb : poln ] ) . as discussed by @xcite ,
the fact that no `` super - polarization '' is seen suggests that there is no strong steady emission from the pulsar .
we confirm that weak but significant right - circular polarization is present for the components near phase zero and phase @xmath55 in fig .
[ fig:2124 ] . in agreement with previous work
, an orthogonal mode transition is seen near phase @xmath50 .
if the pas for emission between phases of @xmath50 and @xmath61 and those for the leading portion of the profile ( to phase @xmath20 , see * ? ? ?
* ) are raised by @xmath57 , then the pa variation approximates an rvm variation for an outer line of sight with the magnetic axis aligned at about @xmath62 to the rotation axis .
such a nearly aligned system may explain the fact that we see emission over most if not all of the pulse period .
this pulsar has been studied at 659 mhz and 1331 mhz by @xcite and at 1373 mhz by @xcite .
the results shown in fig . [ fig:2129 ]
are in agreement with these earlier results , but show more detail .
all observations detect the postcursor and a weaker leading shelf of emission , but the present results show that this extends to at least phase @xmath61 , giving the pulse an overall width of at least 0.4 in phase .
the pa mostly decreases through the pulse , but the present results clearly show a pa `` glitch '' at the phase of the trailing main - pulse component .
the degree of circular polarization is unusually high , especially for the trailing main - pulse component . on average
it is 23% right - circularly polarized , the second highest in table [ tb : poln ] after psr j1603@xmath187202 . the polarimetry of this relatively strong msp ( mean flux density at 9.3 mjy , third to psrs j0437@xmath184715 and j1939 + 2134 , table [ tb : poln ] ) has been studied by a number of groups @xcite . our results , presented in fig . [ fig:2145 ] , are generally in good agreement with these earlier authors , but show more detail .
both @xcite and @xcite found a weak bridge of emission between the leading and main pulse components , but we do not see a bridge .
the apparent bridge in the earlier results is probably an artifact arising from the 2-bit digitization of the input bandpass employed in these systems .
our results show structure in the leading component with at least two sub - components . in agreement with earlier work ,
this component is essentially 100% linearly polarized and has almost constant pa .
the pa variation in the rest of the profile is very complex with regions of increasing and decreasing pa . there are remarkably close orthogonal transitions ( up and back ) at phases 0.223 and 0.231 with corresponding dips in @xmath26 .
there are also sense changes in @xmath3 at phases 0.197 and 0.215 , but these do not seem to be associated with the orthogonal mode jumps which are slightly later in phase .
many of the ppta pulsars either did not have published rms or the published values were insufficiently accurate for summing our data across frequency to form final profiles . as described in section [ sec : obs ] , the variable ionospheric rm was subtracted from the measured values to give an accurate estimate of the interstellar rm for each pulsar .
table [ tb : rms ] gives our estimates of the interstellar rm together with the galactic coordinates and distance ( estimated from the dm using the @xcite model ) of the pulsar and previously published rm values . in some cases
the previously published values do not have the ionospheric component removed but this would generally be smaller than the quoted uncertainty .
we also give the mean value of the line - of - sight component of the interstellar magnetic field in microgauss weighted by the local electron density along the path ( equation [ eq : bpar ] ) .
negative values correspond to fields directed away from earth .
ccccccc psr & gal .
@xmath63 & gal .
@xmath64 & dist . &
rm ( this work ) & rm ( prev .
publ . ) & @xmath65 + & ( deg ) & ( deg ) & ( kpc ) & ( rad m^-2^ ) & ( rad m^-2^ ) & ( @xmath66 g ) + j0437@xmath184715 & 253.39 & @xmath1841.96 & 0.16 & 0.0 @xmath67 0.4 & 1.5 @xmath67 5 & @xmath180.0 @xmath67 0.19 + j0613@xmath180200 & 210.41 & @xmath189.30 & 1.25 & 9.7 @xmath67 1.1 & 19 @xmath67 14 & 0.31 @xmath67 0.03 + j0711@xmath186830 & 279.53 & @xmath1823.28 & 0.86 & 21.6 @xmath67 3.1 & 67 @xmath67 23 & 1.45 @xmath67 0.21 + j1022 + 1001 & 231.79 & @xmath6851.10 & 0.56 & @xmath180.6 @xmath67 0.5 & & @xmath180.07 @xmath67 0.06 + j1024@xmath180719 & 251.70 & @xmath6840.52 & 0.53 & @xmath188.2 @xmath67 0.8 & & @xmath181.56 @xmath67 0.15
+ + j1045@xmath184509 & 280.85 & @xmath6812.25 & 0.30 & 92.0 @xmath67 1.0 & 82 @xmath67 18 & 1.95 @xmath67 0.02 + j1600@xmath183053 & 344.09 & @xmath6816.45 & 5.00 & @xmath1815.5 @xmath67 1.0 & & @xmath180.36 @xmath67 0.02 + j1603@xmath187202 & 316.63 & @xmath1814.50 & 1.17 & 27.7 @xmath67 0.8 & 20.1 @xmath67 5 & 0.90 @xmath67 0.03 + j1643@xmath181224 & 5.67 & @xmath6821.22 & 0.45 & @xmath18308.1 @xmath67 1.0 & @xmath18263 @xmath67 15&@xmath186.08 @xmath67 0.02 + j1713 + 0747 & 28.75 & @xmath6825.22 & 1.05 & 8.4 @xmath67 0.6 & & 0.65 @xmath67 0.05
+ + j1730@xmath182304 & 3.14 & @xmath686.02 & 0.53 & @xmath187.2 @xmath67 2.2 & & @xmath180.92 @xmath67 0.28 + j1732@xmath185049 & 340.03 & @xmath189.45 & 1.41 & @xmath188.5 @xmath67 6.7 & & @xmath180.18 @xmath67 0.15 + j1744@xmath181134 & 14.79 & @xmath689.18 & 0.42 & @xmath181.6 @xmath67 0.7 & & @xmath180.63 @xmath67 0.27 + j1824@xmath182452 & 7.80 & @xmath185.58 & 4.90 & 77.8 @xmath67 0.6 & 1 @xmath67 12 & 0.80 @xmath67 0.01 + j1857 + 0943 & 42.29 & @xmath683.06 & 0.91 & 16.4@xmath67 3.5 & 53 @xmath67 9 & 1.52 @xmath67 0.32 + + j1909@xmath183744 & 359.73 & @xmath1819.60 & 1.27 & @xmath186.6 @xmath67 0.8 & & @xmath180.78 @xmath67 0.09 + j1939 + 2134 & 57.51 & @xmath180.29 & 8.33 & 6.7 @xmath67 0.6 & @xmath1810 @xmath67 9 & 0.12 @xmath67 0.01 + j2124@xmath183358 & 10.93 & @xmath1845.44 & 0.32 & @xmath185.0 @xmath67 0.9 & 1.2 @xmath67 1 & @xmath181.34 @xmath67 0.24 + j2129@xmath185721 & 338.01 & @xmath1843.57 & 0.53 & 23.5 @xmath67 0.8 & 37.3 @xmath67 2 & 0.91 @xmath67 0.03 + j2145@xmath180750 & 47.78 & @xmath1842.08 & 0.62 & @xmath181.3 @xmath67 0.7 & 12 @xmath67 8 & @xmath180.18 @xmath67 0.1 + references : ( a ) @xcite ; ( b ) @xcite ; ( c ) @xcite ; ( d ) @xcite . in most cases our rm measurements are close to previous measurements within the uncertainties .
however , psr j1824@xmath182452 ( psr b1821 - 24 ) stands out ; in this case , it seems likely that the previously published result was incorrect . for several other pulsars ,
the results differ by more than the combined error .
there are several possible reasons for this .
perhaps the most likely is that unknown systematic errors , most probably in calibration , resulted in underestimation of the rm uncertainties .
alternatively , it is possible that in some cases there has been a real change in rm since the earlier measurements .
continuing ppta observations will help to establish if such variations exist .
we have presented new and improved polarization profiles for the 20 msps observed as part of the parkes pulsar timing array project .
because of the frequent and relatively long observations needed by the ppta project , our polarization profiles generally have very high s / n ratios compared to earlier results .
also , we have employed improved signal processing procedures , including corrections for feed cross - coupling and ionospheric faraday rotation . as a result , we have not only defined the polarization properties more accurately , but also revealed previously unknown profile features in many of the pulsars . in a number of cases ,
the newly detected profile features greatly extend the range of pulse phase over which emission is detected .
for example , in psr j1045@xmath184509 , the detection of a low - level leading component and bridge emission joining it to the main pulse has increased the overall observed width of the pulsed emission by nearly a factor of three , from about 0.25 to about 0.7 in phase .
psr j1939 + 2134 ( psr b1937 + 21 ) is an interesting case in which the newly detected emission changes the pulse profile from having just a main pulse and interpulse separated by close to 0.5 in phase and both relatively narrow , to a wide profile covering about 80% of the pulse period . in some ways ,
the situation is similar to that of the crab pulsar , where pulse components with high linear polarization are detected at phases very different to those of the main pulse and interpulse @xcite .
these pulse components are only seen at high radio frequencies , but the crab pulsar also has a highly polarized precursor pulse which is only seen at low radio frequencies @xcite . some , but not all , of the outlying components detected in this study are highly linearly polarized but those in psr j1939 + 2134 are not .
table [ tb : poln ] gives the overall pulse width , that is , the total longitude range over which significant pulsed emission is seen , for each of the ppta pulsars . the distribution of these widths is plotted in fig .
[ fg : widths ] .
only seven of the 20 pulsars have emission spanning less than half of the period and five have emission over more than three - quarters of the period . only in a few of the pulsars
does the pulse profile have the form of a main pulse and interpulse separated by close to half the period . even when there are strong pulse components that can be labelled in this way , there is generally emission at other pulse phases as well .
most normal ( non - millisecond ) pulsars have mean pulse profiles that can be described within the core
cone framework with one to five clearly identifiable pulse components or peaks within a relatively narrow longitude range , typically about @xmath69 .
although the relative amplitude of the various components varies widely , leading to the notion of partial cones @xcite , there often appears to be a symmetry about a central phase , leading to the idea of nested cones ( e.g. , * ? ? ?
* ; * ? ? ?
however , observations with high s / n ratio often show a more complex profile structure with multiple components ( e.g. , psr b0740@xmath1828 in * ? ? ? * ) which are not easily accommodated in the nested - cone model .
this and the wide variation of component amplitudes led to the idea of `` patchy '' emission beams @xcite . pulsars with interpulses and/or wide profiles form another group of multiple component pulsars . only a few normal pulsars have these features , with less than 2% showing interpulse emission @xcite and most of this 2% have short periods and high spin - down luminosities .
many have the main pulse and interpulse separated by close to @xmath58 , but some have smaller spacings and/or bridges of emission between the two components .
good examples are the crab pulsar as discussed above , psr b0950 + 08 and psr b1929 + 10 @xcite
. some interpulse pulsars also have weaker pulse components , often highly linearly polarized , separated from the main pulse and interpulse and usually termed `` precursors '' or `` postcursors '' .
examples are the crab pulsar again , psr b0823 + 26 @xcite and psr b1822 - 09 @xcite . as mentioned above , wide profiles and interpulses are much more common in msps , with more than 60% of the present sample showing these features .
psr j1045@xmath184509 ( fig .
[ fig:1045 ] ) is a good example of a nominal interpulse joined to the main pulse by a bridge of emission .
highly polarized precursors and postcursors are observed in psr j1024@xmath180719 ( fig .
[ fig:1024 ] ) , psr j1603@xmath187202 ( fig . [ fig:1603 ] ) , psr j1713 + 0747 ( fig .
[ fig:1713 ] ) and psr j1730@xmath182304 ( fig .
[ fig:1730 ] ) .
multiple components are ubiquitous in msps .
some components are obvious as clearly resolved peaks in the pulse profile , but others are just points of inflection in the profile and probably some are not obvious in the profile at all .
table [ tb : poln ] gives an estimate of the minimum number of identifiable components in the ppta msps based on peaks or points of inflection in the pulse profiles . in most cases ,
the observed pa variations across these wide profiles are incredibly complex , even after allowance for orthogonal mode transitions , and are not even approximately described by the rvm .
exceptions are psrs j1024@xmath180719 , j1600@xmath183053 and j1732@xmath185049 which have relatively flat pa profiles , and psrs j1022 + 1001 , j1744@xmath181134 and j2124@xmath183358 which have reasonably smooth pa variations that are consistent with the rvm . except for psr j1022 + 1001
, rvm fits to these pa variations imply large impact parameters and/or nearly aligned magnetic and rotational axes . for psr j2124@xmath183358
, a nearly aligned model may well be the correct interpretation since it has emission across essentially the whole period , but in other cases it is not so clear . for most of the pulsars observed , the pa variations are complex with disconnected pa segments , often with opposite slopes in neighbouring regions ( e.g. , psr j0613@xmath180200 , fig .
[ fig:0613 ] ; psr j1603@xmath187202 , fig .
[ fig:1603 ] ; psr j1730@xmath182304 , fig .
[ fig:1730 ] ; psr j2145@xmath180750 , fig .
[ fig:2145 ] ) and non - orthogonal pa jumps ( e.g. , psr j0437@xmath184715 , fig . [ fig:0437 ] ; psr j0711@xmath186830 , fig .
[ fig:0711 ] ; psr 1713 + 0747 , fig .
[ fig:1713 ] ) . within a single component ,
the pa variation is usually continuous and monotonic ( orthogonal jumps excepted ) .
these results suggest that overlapping components may originate in different locations within the magnetosphere . supporting evidence for quasi - independent emission regions comes from phase - resolved spectral index plots which show different spectral indices for different components ( e.g. , * ? ? ?
@xcite have also suggested the idea of independent emission from different parts of the pulsar magnetosphere , in their case , from thin plasma streams emitting coherent curvature radiation . in this model ,
the symmetry axis for pa variations is not the magnetic axis ( as in the dipole polar - cap model ) but the plasma - stream direction .
this could certainly account for the non - rvm pa variations we observe .
however , we do not find evidence for the common occurence of the double features they identify with the basic curvature radiation beam .
the very wide msp profiles we observe ( fig .
[ fg : widths ] ) imply large emission heights in the simple polar - cap model .
in fact , for a pulsar period of 4 ms and an overall pulse width of @xmath70 ( about the median values for our sample ) , equation [ eq : beam ] implies @xmath71 .
one could invoke small inclination angles , but it is more likely that the equation is simply not applicable .
similarly , the standard methods for estimating emission heights can not be applied to msps .
they do not exhibit radius - frequency mapping of component separations and hence that method @xcite can not be used .
likwise , the method of @xcite , which relies on identifying a relativistic aberration / retardation shift between the pulse profile and the pa variations , can not be applied because of the general lack of symmetry about a central longitude for one or both properties . for example , psrs j0437@xmath184715 and j2145@xmath180750 both have reasonably symmetric pulse profiles , but in both cases it is impossible to determine any symmetry centre for the pa variations .
relativistic methods which rely on a height difference between core and conal components @xcite also can not easily be applied to msps since it is generally impossible to uniquely identify components as either core or cone . based on associations with gamma - ray detections , @xcite and
@xcite argued that radio beams from young and millisecond pulsars are emitted from the outer magnetosphere and that caustic effects are important in defining the observed pulse profile . by analogy with gamma - ray pulse profiles ( e.g. , * ? ? ?
* ) such effects may account for the broad frequency - independent pulse profiles observed in msps , including interpulses and connecting bridges .
the results presented in this paper support this idea .
the observed pulse profiles are too wide and too complex to be simply accounted for as emission tangential to the open field lines at low altitudes in the pulsar magnetosphere . observed pa variations are also generally inconsistent with this interpretation .
for example , the extended low - level emission observed for psr j1939 + 2134 ( fig .
[ fig:1939 ] ) shows that main pulse and interpulse in this pulsar are not simple polar beams from opposite magnetic poles as was originally thought . in this context , it is important to note that , for both the crab pulsar and psr j1939 + 2134 , giant radio pulses occur at phases closely related to the phase of the high - energy emission and to the main radio peaks @xcite also supporting the idea of the radio emission originating in the outer magnetosphere .
@xcite discussed the caustic deformation of radio pulse profiles by aberration and retardation , albeit to a more modest degree than that proposed above . trailing sides of profiles
are compressed in width and enhanced in amplitude whereas leading parts are spread out and reduced in amplitude .
they apply these ideas to the main pulse of the msp j1012 + 5307 , which has a strong trailing component and a weaker more extended leading part , and can model the observed profile with a patchy underlying emission and a small relativistic deformation , with the emission originating between the stellar surface and twice that radius .
of the pulsars we observed , only psr j1022 + 1001 and maybe psr j1600@xmath183035 have similar profiles with a relatively strong and narrow trailing component .
it is clear that these ideas are not generally applicable to msps .
@xcite suggested that emission from many young pulsars arises from the outermost open field lines at relatively high altitudes , between 5 and 15 per cent of the light - cylinder radius .
we are proposing that for msps ( and some young pulsars as well ) the emission zone extends much higher , up to or even beyond the null - charge surface , with caustic deformations a dominant influence on the profile morphology . to account for the multiple components and complex pa variations observed in msps
, much structure is required in the underlying emission .
this could originate from isolated plasma streams as suggested by @xcite with the observed pa determined by the projection of the stream axis or by the local magnetic field direction . in the outer magnetosphere
, such currents will significantly deform the basic dipolar field , leading to the observed small - scale pa variations and discontinuities .
it is worth noting though that the observed complex profile and pa variations are extremely stable over timescales of decades ( compare the psr j0437@xmath184715 profiles in @xcite and in fig .
[ fig:0437 ] ) .
observations of nulling and mode - changing in normal pulsars ( e.g. , * ? ? ?
* ; * ? ? ?
* ) show that magnetospheric currents can vary on short timescales , but to date there is little evidence that such phenomena exist in msps . even in normal pulsars ,
the properties of a given mode seem quite stable .
inwardly directed emission beams from charges flowing toward the neutron star @xcite provides another possible explanation for the multiple and overlapping components of msp pulse profiles . despite the generally more complex pulse profiles of msps compared to normal pulsars , there are many similarities in the emission characteristics .
fractional linear polarizations are often very high for both classes of pulsar , orthogonal - mode transitions are ubiquitous in both classes and circular polarization is often seen , but is almost always weaker than the linear polarization .
truly orthogonal polarization transitions ( both circular and linear changing sign ) do seem more common in msps with at least five clear examples ( in psrs j0437@xmath184715 , j1600@xmath183053 , two in j1643@xmath181224 and j1713 + 0747 ) among the 20 ppta pulsars presented here .
overall though , the similarities in polarization behaviour outweigh the differences , strongly suggesting that the basic radio emission mechanism is the same for both normal pulsars and msps .
we derive rms for all 20 pulsars , eight of which have no previously published rms .
for those with previously published rms , our measurements have smaller estimated uncertainties and we believe they are more accurate than the previous values . for three pulsars ( j1824@xmath182452 ,
j1939 + 2134 and j2145@xmath180750 ) our measurements differ significantly , indicating systematic errors , most probably in the earlier measurements .
rms and implied mean values of the line - of - sight component of the interstellar magnetic field are small for high - latitude and more distant pulsars in accord with earlier results ( e.g. , * ? ? ?
most measured values are less than or about @xmath72 g , but that for psr j1643@xmath181224 at @xmath73 g is an exception .
the estimated distance for this pulsar is only about 0.5 kpc and the path to it traverses the north polar spur @xcite .
the large mean field strength almost certainly results from compressed interstellar fields in the shell(s ) which form this prominent feature of the galactic radio background . the parkes pulsar timing array ( ppta ) project is a collaboration between a number of groups both in australia in other countries established with the support of rnm s australian research council federation fellowship ( # ff0348478 ) .
gh is the recipient of an australian research council qeii fellowship ( # dp0878388 ) .
wmy is supported by nsfc project 10673021 , the knowledge innovation program of the chinese academy of sciences , grant no .
kjcx2-yw - t09 and national basic research program of china ( 973 program 2009cb824800 ) .
we thank members of the ppta collaboration who assisted with the observations reported in this paper .
we thank andrew gray and ken tapping for making the penticton ionospheric modelling program available to us .
the parkes radio telescope is part of the australia telescope , which is funded by the commonwealth of australia for operation as a national facility managed by the commonwealth scientific and industrial research organisation .
manchester r. n. , 2008 , in bassa c. , wang z. , cumming a. , kaspi v. m. , eds , 40 years of pulsars : millisecond pulsars , magnetars and more vol .
983 , the parkes pulsar timing array project .
american institute of physics , new york , pp 584592 | polarization profiles are presented for 20 millisecond pulsars that are being observed as part of the parkes pulsar timing array project .
the observations used the parkes multibeam receiver with a central frequency of 1369 mhz and the parkes digital filterbank pulsar signal - processing system pdfb2 .
because of the large total observing time , the summed polarization profiles have very high signal / noise ratios and show many previously undetected profile features .
thirteen of the 20 pulsars show emission over more than half of the pulse period .
polarization variations across the profiles are complex and the observed position angle variations are generally not in accord with the rotating - vector model for pulsar polarization .
never - the - less , the polarization properties are broadly similar to those of normal ( non - millisecond ) pulsars , suggesting that the basic radio emission mechanism is the same in both classes of pulsar .
the results support the idea that radio emission from millisecond pulsars originates high in the pulsar magnetosphere , probably close to the emission regions for high - energy x - ray and gamma - ray emission .
rotation measures were obtained for all 20 pulsars , eight of which had no previously published measurements .
[ firstpage ] pulsars : general polarization radio continuum : stars |
the rapid emergence of the field of spintronics ( also known as magnetoelectronics @xcite ) brought to the center of scientific research the so - called half - metallic ferromagnets ( like heusler alloys @xcite or some oxides @xcite ) .
these compounds present metallic behavior for one spin - band while they are semiconducting or insulators for the other spin - band , resulting to perfect spin - polarization , at least for the bulk , at the fermi level . for realistic application ferromagnets create large stray fields and thus lead to considerable undesirable energy losses .
thus , to this respect the case of half - metallic ferrimagnets is more interesting .
there are several ways to create a half - metallic ferrimagnet , either by doping a semiconductor like fevsb @xcite or due to the simultaneous presence of mn and another transition - metal atom in the case of alloys with small total moment due to the small number of valence electrons ( e.g. femnsb @xcite or mn@xmath0val @xcite ) .
recently akai and ogura have proposed another route to fully - compensated half - metallic ferrimagnetism based on the doping of diluted magnetic semiconductors @xcite . except heusler and oxides , also transition - metal chalcogenides like cras and pnictides like crse
are known to present half - metallic ferromagnetism when they crystallize in the metastable zinc - blende structure .
the first experimental evidence was provided in the case of cras thin - films by the group of akinaga in 2000 @xcite and many more experiments have confirmed these results @xcite .
experiments agree with prediction of ab - initio calculations performed by several groups @xcite . in the case of the half - metallic ferromagnets like cras or crse
, the gap in the minority - spin band arises from the hybridization between the p - states of the @xmath1 atom and the triple - degenerated @xmath2 states of the transition - metal and as a result the total spin - moment , @xmath3 , follows the slater - pauling ( sp ) behavior being equal in @xmath4 to @xmath5 where @xmath6 the total number of valence electrons in the unit cell @xcite .
recently theoretical works have appeared attacking also some crucial aspects of these alloys like the exchange bias in ferro-/antiferromagnetic interfaces @xcite , the stability of the zinc - blende structure @xcite , the dynamical correlations @xcite , the interfaces with semiconductors @xcite , the exchange interaction @xcite and the temperature effects @xcite . in a recent publication @xcite
, we have investigated the properties of several compounds ( cras , crsb , crse , crte , vas and mnas ) when we create an excess of the transition metal atoms ; transition metal atoms occupy also antisites occupied in the perfect compound by the @xmath1 atom . to perform this study we used the full potential nonorthogonal local orbital minimum basis band structure scheme ( fplo ) and we simulated disorder via the coherent potential approximation @xcite ; we also used the theoretical equilibrium lattice constants for which the perfect compounds are half - metallic ferromagnets @xcite .
the most interesting case was the cr - based compounds .
the cr - impurities in the case of antisites at the sublattice occupied by the @xmath1 atoms couple antiferromagnetically to the existing cr atoms at the ideal sites and destroy ferromagnetism .
but these compounds stay half - metallic for large concentration of antisites exhibiting half - metallic ferrimagnetism .
also in v@xmath7as@xmath8 compound the v atoms at the antisites couple antiferromagentically to the v atoms at the perfect sites but the exchange splitting of the v impurities is small and the fermi level cross the states of these atoms . in this manuscript
we expand this study to cover also the case where at the antisites we find transition metal atoms of different chemical type than the perfect sites .
more precisely we studied the case of mn impurites in cr[as@xmath9mn@xmath10 and cr[se@xmath9mn@xmath10 and the case of cr and mn impurities in v[as@xmath9cr@xmath10 and v[as@xmath9mn@xmath10 compounds .
we find that in the case of cr - compounds the half - metallic ferrimagnetic behavior is again present , and the same is also true for the v[as@xmath9(cr or mn)@xmath10 compounds since cr and mn atoms couple antiferomagnetically to the v atoms at the perfect site and have a larger exchange splitting than the v impurities in v@xmath7as@xmath8 . for the studied t[z@xmath8t@xmath11 $ ] compounds.[fig1 ] ]
before going in details of the density of states ( dos ) or atomic spin moments we will present our results on the total spin moments . in figure [ fig1 ] we have included the different cases under study going up to x=0.5 where half of the @xmath1 atoms have been substituted by transition metal atoms . of course such
a case is unlikely to be stable in the zinc - blende structure but we have gone so far to make physics more transparent.the perfect cras compound has a total spin moment in the unit cell of 3 @xmath4 since there are 11 valence electrons , crse with one valence electron more in the unit cell has a spin moment of 4 @xmath4 and vas with 10 valence electrons a spin moment of 2 @xmath4 . in all cases the total spin moment decreases with the concentration , a clear sign of the antiferromagnetic coupling between the transition metal atoms at different sites . in the case of cras , substituting cr or mn for as results to identical behavior ( solid black lines with open diamonds and stars , respectively in the figure ) and the total spin moment is almost independent of the chemical type of the transition - metal atom at the antisite .
this is also true to a lesser extent for crse ( red open triangles for cr[se@xmath8cr@xmath10 and blue triangles for cr[se@xmath8mn@xmath10 ) .
for the same concentration x the compound with mn has a slightly larger total spin moment and as the concentration increases this difference also increases . in the case of vas
the differences are more pronounced .
when we substitute v for as , v[as@xmath8v@xmath10 , the v atoms at the ideal and the antisites are antiferromagnetically coupled .
as the concentration increases the absolute value of the v spin moments decreases and for x=0.5 the total spin moment is almost zero and the v atoms are almost non - magnetic . cr and mn impurities , on the other hand , lead to larger total spin moment and the resulting compounds are ferrimagnets with strong magnetic elements .
overall the v[as@xmath8mn@xmath10 compounds have a larger spin moment than the v[as@xmath8cr@xmath10 ones .
we should note that although the behavior of the total spin moments reveals the existence of ferrimagnetism it does not give any information about the half - metallic character and we should look how the dos changes with the concentration @xmath12 for every case to be sure if the half - metallicity is preserved or lost .
.total and atom - resolved spin magnetic moments for the cr[z@xmath8y@xmath13 compounds where z= as or se and y= cr or mn . as `` imp ''
we denote the transition - metal atoms which are located at antisite positions .
note that the atomic moments have been scaled to one atom .
[ cols="<,^,^,^,^,^,^,^,^ " , ] [ table2 ] we still have not discussed the behavior of the impurity atoms in the case of vas . from the table [ table2 ]
it is obvious that , contrary to cr - based compounds , the impurity cr or mn atoms have a spin magnetic moment of around -2 @xmath4 , while the v impurity atoms ( see table 1 in reference @xcite ) have a spin moment of around -1 @xmath4 . to understand this behavior we have to look at the atom - resolved dos in figure [ fig4 ] .
the exchange splitting of the v impurities at the antisites is very small and the majority ( spin - down ) and minority ( spin - up ) double degenerated @xmath14-states are completely occupied ( peaks with the higher intensity located around -1 ev ) . also one of the minority and two of the majority triple - degenerated @xmath2 states are occupied resulting in a spin moment of around -1 @xmath4 and the fermi level is crossing the spin - down band of the @xmath2 states . for cr and mn impurities
the exchange splitting is larger and all three majority ( spin - down ) @xmath2-states are occupied while the picture for the minority states does not change and consecutively the spin moment of the impurity is around -2 @xmath4 .
also this results to a spin - down gap separating the fully occupied d - states from the antibonding states lying higher in energy and thus to half - metallic ferrimagnetism .
the exchange splitting is larger for mn than cr and the spin - down gap is larger in the former case .
as we change the concentration of the impurity atoms their spin moment only slightly changes due to the stronger hybridization with the v atoms at the ideal sites .
finally we should note that as we decrease the number of valence electrons and pass from crse to cras and then to vas the ionicity of the compounds becomes smaller and so do the gaps created by the impurity atoms ( smaller exchanged splitting for the same impurity atom when it is found in environment with smaller ionicity ) .
we have complemented our study presented in a recent publication [ galanakis i et al 2006 _ phys .
rev b _ * 74 * 140408(r ) ] where we have shown that in the case of cras and related transition - metal chalcogenides and pnictides , crystallizing in the zinc - blende structure , the excess of the transition - metal atoms leads to half - metallic ferrimagnetism . in this contribution
we show that there may be additional advantages when the excess of the transition - metal atom is not of the same chemical type as the one at the perfect lattice sites . in cras and crse
, the creation of mn antisites keeps the half - metallic character and the mn impurities are antiferromagnetically coupled to the cr ones leading to half - metallic ferrimagnetism .
the larger exchange splitting of the mn atoms with respect to the cr ones makes the half - metallic ferrimagnetism even more robust since the gap - width is larger with respect to the case of cr antisites . even in vas , which looses its half - metallic character upon creation of v antisites , the appearance of cr or mn antisites
is accompanied with the appearance of half - metallic ferrimagnetism .
these results may suggest a new way to achieve stable half - metallic ferrimagnets , which is crucial for spintronic applications with respect to half - metallic ferromagnets due to the lower stray fields created by these materials .
authors acknowledge the computer support of the `` leibniz institute for solid state and materials research dresden '' .
999 soulen jr r j , byers j m , osofsky m s , nadgorny b , ambrose t , cheng s f , broussard p r , tanaka c t , nowak j , moodera j s , barry q and coey j m d 1998 _ science _ * 282 * 85 ; park j h , vescovo e , kim h j , kwon c , ramesh r and venkatesan t 1998 _ nature _ * 392 * 794 ; parker j s , watts s m , ivanov p g and xiong p 2002 * 88 * 196601 ; kmper k p , schmitt w and gntherodt g 1987 * 59 * 2788 deng j j , zhao j h , bi j f , niu z c , yang f h , wu x g , and zheng h z 2006 _ j. appl . phys .
_ * 99 * 093902 zhao j h , matsukura f , takamura k , chiba d , ohno y , ohtani k , and ohno h 2003 _ mat .
proc . _ * 6 * 507 leai m , mavropoulos p , enkovaara j , bihlmayer g , and blgel s 2006 * 97 * 026404 ; chioncel l , katsnelson m i , de wijs g a , de groot r a , and lichtenstein a i 2005 b * 71 * 085111 ; skomski r and dowben pa 2002 _ europhys . | in a recent publication [ galanakis i et al 2006 b * 74 * 140408(r ) ] we have shown that in the case of cras and related transition - metal chalcogenides and pnictides , crystallizing in the zinc - blende structure , the excess of the transition - metal atoms leads to half - metallic ferrimagnetism .
the latter property is crucial for spintronic applications with respect to ferromagnets due to the lower stray fields created by these materials .
we extend this study to cover the case where the transition - metal atoms sitting at antisites are not identical to the ones in the perfect sites . in cr - based compounds
, the creation of mn antisites keeps the half - metallic ferrimagnetic character produced also by the cr antisites . in the case of vas , cr and mn antisites keep the half - metallic character of vas ( contrary to v antisites ) due to the larger exchange - splitting exhibited by these atoms . |
this work was supported in part by the national natural science foundation of china , and the education ministry of china .
k@xmath71hn , j. kaplan and e.g.o .
safiani , nucl . phys . *
b157 * , 125 ( 1979 ) ; b. guberina , j.h .
k@xmath72hn , r.d .
peccei and r. r@xmath71ckl , nucl . phys . * b174 * , 317(1980 ) ; p. cho and a.k .
. rev . * d53 * , 150(1996 ) . | we study the production of @xmath0 in @xmath1 annihilation through two virtual photons .
the cross section is estimated to be 23 fb at @xmath2 gev , which is smaller by a factor of six than the calculated cross section for the same process but through one virtual photon . as a result , while the annihilation into two photons may be important for certain exclusive production processes , the big gap between the inclusive production cross section @xmath3 0.9pb observed by belle and the current nonrelativistic qcd prediction of @xmath40.15pb still remains very puzzling .
we find , however , as the center - of - mass energy increases ( @xmath5 gev ) the production through two virtual photons @xmath6 will prevail over that through one virtual photon , because in the former process the photon fragmentation into @xmath7 and into the charmed quark pair becomes more important at higher energies .
+ pacs number(s ) : 12.40.nn , 13.85.ni , 14.40.gx production of @xmath8 through two photons in @xmath1 annihilation + kui - yong liu and zhi - guo he + department of physics , peking university , beijing 100871 , people s republic of china + kuang - ta chao + china center of advanced science and technology ( world laboratory ) , beijing 100080 , people s republic of china and department of physics , peking university , beijing 100871 , people s republic of china charmonium production is one of the important processes to test quantum chromodynamics ( qcd ) both perturbatively and non - perturbatively .
because of the simpler parton structure involved , which will reduce the theoretical uncertainty , charmonium production in @xmath1 annihilation is expected to be more decisive in clarifying the production mechanisms of heavy quarkonia .
the two b factories with babar and belle are collecting huge data samples of continuum @xmath1 annihilation events , which will allow us to have a fine data analysis for charmonium production .
recently , charmonium production in @xmath1 annihilation has become more interesting and puzzling , because of the large gap between the belle measurements@xcite and the theoretical calculations for both inclusice@xcite and exclusive @xcite charmonium production via double @xmath9 pairs based on nonrelativistic qcd ( nrqcd ) .
for the inclusive processes , belle has reported a measurement on the @xmath7 production in @xmath1 annihilation at @xmath2 gev@xcite , and found that a very large fraction of the produced @xmath7 is due to the double @xmath9 production in @xmath1 annihilation@xcite @xmath10 which corresponds to @xcite @xmath11 in contrast , the predicted values for the cross section in nrqcd ( the color - octet contribution is negligible for this process ) are much smaller than the data@xcite . in a recent analytical calculation and numerical estimation for the inclusive charmonia production including all s - wave , p - wave and d - wave states via double @xmath9 in nrqcd@xcite ,
we find @xmath12 this value is consistent with other previous calculations , including those obtained based on the quark - hadron duality hypothesis@xcite , but smaller than the belle data by about a factor of six . for the exclusive processes , the belle measurement@xcite for the cross section of @xmath13 is about an order of magnitude larger than the nrqcd calculation for @xmath14@xcite .
in order to solve the problem , calculations for the exclusive double - charmonium production from @xmath1 annihilation into two virtual photons are performed@xcite , and it is pointed out that the cross section for @xmath15 is larger than that for @xmath16 by about a factor of 3.7 , despite of possible uncertainties due to the choice of input parameters@xcite . this is a interesting result since it indicates that the @xmath1 annihilation through two virtual photon fragmentation may make important contributions to certain processes , and it might substantially reduce the discrepancy between experiment and theory for the exclusive process @xmath17 ( note that it is essential to check experimentally whether @xmath18 is largely misidentified with @xmath13 ) .
in this situation , it is necessary to examine the contribution of @xmath1 annihilation through two virtual photons to the inclusive production of @xmath7 . in the following
we will calculate the complete o@xmath19 color - singlet inclusive cross section for @xmath20 , and compare the production rate through two virtual photons with that through one virtual photon , to see whether the annihilation through two virtual photons can decrease the discrepancy between the belle measurement on @xmath21 production and the calculations based on nrqcd .
.,width=529,height=604 ] following the nrqcd factorization formalism , color - singlet scattering amplitude of the process @xmath22 in fig .
[ feynman ] is given by @xmath23 where @xmath24 is the intermediate @xmath9 pair which is produced at short distances and then evolves into a specific charmonium at long distances .
@xmath25 , @xmath26 , @xmath27 are respectively the color - su(3 ) , spin - su(2 ) , and angular momentum clebsch - gordon coefficients for @xmath28 pairs projecting out appropriate bound states . for @xmath7
the coefficient @xmath29 can be related to the radial wave function of the bound state and reads @xmath30 the spin projection operator can be defined as@xcite @xmath31 we write the spin projection operator which will be used in the calculation as @xmath32 where @xmath33 is the mass of the charmonium , which equals to @xmath34 in the nonrelativistic approximation .
( solid line ) and @xmath35 ( dotted line ) as functions of the scattering angle of @xmath7.,width=453,height=377 ] the amplitude for the upper feynman graph in fig .
[ feynman ] can be written as @xmath36\bar{u}(p_3)\gamma_{\mu}v(p_4 ) \frac{1}{p^2(p_3+p_4)^2},\end{aligned}\ ] ] and the amplitude for the corresponding flipped graph is denoted as @xmath37 .
the amplitude for the lower feynman graph in fig . [ feynman ] reads @xmath38 and the amplitude for the flipped graph is denoted as @xmath39 .
the calculation of cross section for @xmath40 is straightforward .
the differential cross section can be written in the form @xmath41 where @xmath42 represents the elements of the quadruple integral related to the final state phase space ( see , e.g. ref .
@xcite ) , and @xmath43 is the unpolarized module squared of the amplitude . for simplicity we will not write down their lengthy expressions here
. with eq .
( [ cross ] ) we can evaluate the inclusive cross sections for @xmath7 production from @xmath1 through two virtual photons .
the input parameters used in the numerical calculations are the same as in ref.@xcite @xmath44 @xmath45 now we give the numerical result at the belle energy @xmath2 gev : @xmath46 it is well known that @xmath47 is a pure electromagnetic process ( except for the hadronization of the quark pair at long distances ) , while @xmath48 involves both electromagnetic and strong interactions . for a naive order of magnitude estimate , the ratio of the production rate of the former to the latter would be proportional to @xmath49 , but the photon fragmentation into @xmath7 and into the charmed quark pair can substantially enhance the former . to see the role of photon fragmentation , we plot the differential cross sections of @xmath50 and @xmath51 as functions of the scattering angle of @xmath7 at @xmath2gev in fig .
[ distribution ] .
( here for the one - photon process we choose @xmath520.26 . )
one can see that the small angle @xmath7 production in @xmath53 is dominant .
this indicates that most of the @xmath7 production comes from the photon fragmentation ( corresponding to the upper graph in fig .
[ feynman ] with one photon fragmenting to @xmath7 and the other fragmenting to a charm quark pair ) .
indeed , our calculation shows that at @xmath2gev , the contribution from fragmentation graphes is about seventy - two percent of the total cross section . in fig .
[ ratio ] we show the ratio of the fragmentation contribution to the total cross section as a function of the @xmath1 center - of - mass energy in @xmath54 .
it is clear that the photon fragmentation becomes more and more dominant as the center - of - mass energy increases .
this is in agreement with the observation in the @xmath55 exclusive production through two virtual photons@xcite . from the above discussions we have seen the importance of the photon fragmentation to @xmath7 as well as to the charm
quark pair in the two - photon process @xmath50 .
a even more crucial result is that at high @xmath1 center - of - mass energies the contribution through two virtual photons will prevail over that through one virtual photon in the production of @xmath21 .
the reason lies simply in the fact that the virtuality of the photon in the two - photon process can be as small as @xmath56 , whereas it is as large as @xmath57 , the center - of - mass energy squared in the one - photon process . in fig .
[ num ] we show the cross sections of @xmath54 ( solid line ) and @xmath58 ( dotted line ) as functions of the center - of - mass energy @xmath59 .
we see clearly that the cross section for @xmath54 decreases very slowly , whereas that for @xmath60 decreases rapidly as @xmath59 increases . at @xmath61 gev ,
the two photon process becomes to prevail over the one photon process .
however , unfortunately , at the belle energy @xmath2 gev , since the enhancement effect due to the factor @xmath62 is not large enough as compared with the suppression factor @xmath49 , we find @xmath63 fb , which is still much smaller than @xmath64 fb@xcite , and therefore is negligible .
( solid line ) and @xmath60 ( dotted line ) as functions of the center - of - mass energy , width=453,height=377 ] in summary , we have calculated the complete @xmath65 color - singlet inclusive cross sections for @xmath21 production from @xmath66 annihilation into two photons . due to the suppression factor of @xmath49 ,
at the @xmath1 center - of - mass energy @xmath2 gev the cross section of this process is smaller by about a factor of six than that from @xmath66 annihilation into one photon .
we then conclude that while the @xmath66 annihilation into two photons could be helpful in solving the puzzle for the exclusive @xmath67 production , it can do very little to reduce the big gap between the observed inclusive production cross section of @xmath68 pb and the current nrqcd predictions of about 0.15 pb .
this puzzle still needs to be explained with new theoretical considerations .
we find , however , as the center - of - mass energy increases(@xmath6920 gev ) the production through two photons @xmath50 will prevail over that through one photon , because in the former case the photon fragmentation into @xmath7 and into the charmed quark pair becomes more important at higher energies . |
when a fluctuating field is confined by material bodies , effective forces arise on them . this is due to the fact that the bodies impose boundary conditions on the medium , depending on their geometry , mutual position and material properties , which leads to a modification of the allowed fluctuations in the medium .
the last leads to a dependence of the ground state , or the thermodynamic potential of the system ( say the free energy ) on the geometry of the system and on the distances between its ( macroscopic ) components . in order to change these distances
one has to apply a force that depends on the induced change of the allowed fluctuations .
if the fluctuations are long - ranged the corresponding forces are also long - ranged .
the existence of such long - ranged fluctuation mediated forces is called the casimir effect and the corresponding forces
casimir like forces @xcite , after the dutch physicist hendrik casimir who in 1948 predicted an attractive force between two parallel perfectly conducting metal plates @xcite separated by a finite gap @xmath0 in vacuum at zero temperature . in order
the force to be long - ranged ( i.e. , to decay in a power - law and not in an exponential with the distance way ) the interactions in the system have to be mediated by massless excitations
photons , goldstone bosons , acoustic phonons , etc . considered in this general form
the casimir effect is a subject of investigations in condensed matter physics , quantum electrodynamics , quantum chromodynamics and cosmology .
the results are summarized in an impressive number of reviews @xcite . when the fluctuating field is the electromagnetic one the effect is known as the quantum _ electro - dynamical _ ( qed ) casimir effect . there
the casimir force is caused by zero - point and thermal fluctuations of the electromagnetic field . in first approximation
, it depends only on the velocity of light @xmath7 , planck s constant @xmath8 , the temperature @xmath3 , and the separation distance between the bodies @xmath0 , i.e. , this force to a great extend is _
universal_. a more advanced theory , the so - called lifshitz theory , reveals the dependence of the casimir force on the material properties of the bodies @xcite and geometry of their boundary surfaces @xcite .
thirty years after casimir s prediction , m. e. fisher and p. g. de gennes suggested that the fluctuating medium , confined between the bodies can be a fluid , the fluctuating field being the field of its order parameter , in which the interactions in the system are mediated not by photons but by different type of massless excitations like critical fluctuations or goldstone bosons ( spin waves ) .
the corresponding casimir effect is known as the _ thermodynamic _ casimir effect @xcite . when the confined fluid approaches it s critical point , the corresponding fluctuations are the critical fluctuations of the order parameter and then the effect is usually called _
critical _ casimir effect . in first approximation
, the thermodynamic casimir effect depends only on the gross features of the system its dimensionality @xmath9 and the symmetry of the ordered state @xmath10 ( both defining the so - called bulk universality class of the system ) and on the boundary conditions ( determined by the surface universality classes ) .
therefore , to a great extend the thermodynamic casimir force is also _
universal_. so far the critical casimir effect has enjoyed two general reviews @xcite and some concerning specific aspects of it @xcite . currently the casimir effect is an object of intensive studies both in its original formulation due to casimir as well as in its thermodynamic manifestation . in the present article
we will report theoretical results dealing with the critical casimir effect .
let us note , that the critical casimir effect has been already directly observed , utilizing light scattering measurements , in the interaction of a colloid spherical particle with a plate @xcite both of which are immersed in a binary liquid mixture .
the effect has been also studied in @xmath11he @xcite,@xcite , as well as in @xmath12he@xmath11he mixtures @xcite in the context of forces that determine the properties of a film of a substance in the vicinity of its bulk critical point . in ref .
@xcite and ref .
@xcite one has performed measurements of the casimir force in thin wetting films of binary liquid mixture .
one the theoretical side , the effect has been studied via exact calculations in the two - dimensional ising model @xcite , the three dimensional spherical model @xcite , via conformal - theoretical methods @xcite , within mean - field type calculations on ising type @xcite and @xmath13 models @xcite , through renormalization - group studies via @xmath14-expansion @xcite and via fixed dimension @xmath9 techniques @xcite of @xmath15 models , as well as via monte - carlo calculations @xcite .
the fluctuation of importance in all of the above mentioned models are of thermal origin since all these models possess non - zero critical temperature . in some systems , however , certain quantum parameters govern the fluctuations near their critical point which is usually close to or at the zero temperature @xcite . in this particular case one
speaks of a _ quantum critical _ casimir effect @xcite .
the rapid progress in nanotechnology has resulted in the growth of interest in fluctuation - induced phenomena , which play a dominant role between neutral non - magnetic objects at short separation distances ( below a micrometer ) .
the van der waals and casimir forces , both known under the generic name _
dispersion forces _ , play a key role in micro- and nano - electromechanical systems ( mems / nems ) @xcite operating at such distances . indeed ,
upon scaling down devices , the dispersion forces can induce some usually undesirable non - linear behaviors in such systems @xcite .
irreversible phenomena appear such as stiction ( i.e. , irreversible adhesion ) or pull - in due to mechanical instabilities @xcite .
therefore the ability to modify the casimir interaction can strongly influence the development of mems / nems .
several theorems seriously limit , however , the possible search of repulsive casimir forces @xcite .
currently , apart from some suggestions for achieving casimir repulsion in systems out of equilibrium @xcite , the only experimentally well verified way to obtain repulsive casimir force is to have interaction between two different materials characterized by dielectric permittivities @xmath16 and @xmath17 such that @xcite @xmath18 along the imaginary frequency axis , with @xmath19 being the dielectric permittivity of the medium in between them . in refs .
@xcite casimir repulsion was indeed observed experimentally for the sphere - plate geometry . in the current article
we study the interplay between the critical casimir force and the van der waals one in a system composed out of two flat parallel slabs both immersed in a critical fluid .
let us note that both the critical casimir and van der waals forces are fluctuation induced ones but due to the fluctuations of different entities . for terminological clarity , let us also remind that in colloid sciences fluid mediated interactions between two surfaces or large particles are usually referred to as solvation forces @xcite .
thus , we study here a particular case of such a force when the fluid is near its critical point . in our system
we suppose that the slabs are coated by thin layers of some substances , confining either a non - polar one - component fluid or a non - polar binary liquid mixture . we suppose that the liquid phase of the one - component fluid or one of the components of the binary liquid mixture are strongly adsorbed by _ both _ coating layers , i.e.
, they ensure the so - called @xmath1 boundary conditions . the slabs , on the other hand , influence the fluid by long - range competing dispersion potentials .
in the case of a simple fluid these potentials increase the adsorption of the fluid near one of the surfaces , leading to preference there of its liquid phase , and decrease it near the other one . in the case of a binary fluid mixture
the substrates prefer one of the components near the top and the other one near the bottom of the system .
we will demonstrate that this experimentally realizable competition between the effects due to the coatings and the slabs can result in interesting effects like sign change of the casimir force , acting between the surfaces confining the fluid when one changes @xmath20 , @xmath4 or @xmath0 .
the last facts can potentially be used in designing nanodevices and for governing the behavior of objects at small , below micrometer , distances . if a fluid system possesses a surface it breaks the spatial symmetry of the bulk system .
the quantitative effects of the presence of a surface on the thermodynamic behavior of the system depends on the penetration depth of this symmetry breaking effect into the volume .
there are two phenomena which increase the surface effects : long - range interactions and long - range correlations .
they can act separately , or simultaneously , which leads to an interesting interplay of the effects due to any of them @xcite . the penetration depth due to the correlations
is set by the correlation length @xmath21 of the order parameter of the system ; @xmath21 becomes large , and theoretically diverges , in the vicinity of the bulk critical point @xmath22 : @xmath23 , @xmath24 , and @xmath25 , @xmath26 , where @xmath27 and @xmath28 are the usual critical exponents . if the system is made finite , e.g. , by the introduction of a second wall the behavior of the fluid is further enriched .
when @xmath21 becomes comparable to the characteristic system size , say @xmath0 , the size dependence of thermodynamic functions enters through the ratio @xmath29 , i.e. , takes a scaling form given by the finite - size scaling theory @xcite that incorporates , _ inter alia _ , shift of the critical point of the system @xcite . below @xmath30
if the confining walls of the film geometry consist of the same material , one encounters the phenomenon of capillary condensation @xcite where the confinement of the fluid causes , e.g. , the liquid vapor coexistence line to shift away from the coexistence line of the bulk fluid into the one - phase regime . we will demonstrate in the current article that in the envisaged realization of our system depending on the material properties of the slabs the phase diagram of the finite system might essentially differ from that one of the well studied case of a system with short - ranged type interactions and strongly adsorbing surfaces .
the article is arranged as follows . in sec .
[ sec : thermalcas ] we recall and comment on the finite - size behaviour of systems with dispersion forces extending the known facts to the expected behavior of the casimir and the net forces when they act between walls being semi - infinite slabs coated by some thin substances . by doing so , we especially pay attention to the conditions under which the effects stemming from these interactions are relevant .
section [ sec : model ] presents the corresponding lattice gas models suitable for the investigation of fluid media with account of the long - ranged van der waals interactions .
here we identify the main coupling parameters characterizing the interactions in the systems and in sec .
[ sec : finsizebehav ] the equation for the equilibrium profile of the finite - size order parameter is obtained , which we later use to calculate the forces of interest .
section [ sec : results ] presents the numerical results for the behaviour of the investigated forces followed by sec .
[ sec : phasebehaviour ] where the phase behaviour of the considered type fluid system is briefly discussed .
the experimental feasibility of the predicted effects is discussed in sec .
[ sec : experimentalreal ] .
here we also comment on the possible application of our findings in design of nanodevices and for governing the behavior of objects , say colloidal particles , at small distances . the article ends with a summary and discussion section sec .
[ sec : disandconcrem ] .
important technical details concerning the hamaker term for a van der waals system of two different substances separated by a fluid medium are presented in appendix [ sec : hamakerterm ] .
and @xmath31 , coated by thin layers of some other substances @xmath32 and @xmath33 , respectively , confining some fluid medium @xmath34 a non - polar one - component fluid [ as an example we depicted the carbon dioxide molecules @xmath35 @xmath36 or a binary mixture @xmath37 composed out of the molecules of the non - polar liquids @xmath38 and @xmath39 [ the depicted examples include tetrachloroethene @xmath40 as the substance @xmath38 , and benzene @xmath41 as the substance @xmath39 ] . the confined fluid medium is considered embedded on a lattice in which @xmath36 , some nodes are occupied by a particle and others are not thus depicting the `` liquid '' and `` gas '' states respectively at some values of fluid temperature @xmath3 and chemical potential @xmath4 , or @xmath37 some of the nodes are occupied by a molecule from the substance @xmath38 ( the `` liquid '' state ) and the rest are occupied by the molecules belonging to the species @xmath39 ( `` gas '' state ) . the confining walls
impose on the fluid medium boundary conditions of strong adsorption on the coating layers , i.e. , the nearest to the coating substances layers are entirely occupied by the particles of the one - component fluid or if the medium is a binary liquid mixture by the particles of one of its components ( in the presented figure we choose the molecules of the specie @xmath38 ) . ]
let us consider some fluid medium confined between two parallel slabs of some materials @xmath43 and @xmath31 .
any of the slabs is coated by thin solid films of some other substances @xmath32 and @xmath33 , respectively ( see fig .
[ fig : ocf_and_blm ] ) .
let the slabs are situated at some distance @xmath0 from each other .
we suppose that the thicknesses of the coating films are negligible . in the remainder of the text we are going to designate each slab and the thin solid film that coats it as a `` wall '' , and
refer to any of its two components separately only when this is necessary .
if the fluid medium is in contact with a particle reservoir with a chemical potential @xmath4 , the grand canonical potential @xmath44 of this medium in excess to its bulk value @xmath45 depends on @xmath0 and , thus , one can define the effective force @xmath46 per cross sectional area @xmath47 and @xmath48 , due to the fluctuations of the medium and dispersion interactions in it as @xmath49 where the superscript @xmath50 designates the boundary conditions which the confining walls impose on the fluid medium ( see above ) , @xmath51 is the excess grand canonical potential per unit area @xmath47 , @xmath52 is the total grand canonical potential , @xmath53 is the density of the bulk grand canonical potential , and @xmath54 @xcite .
let us stress that , as pointed out in ref .
@xcite , one should keep in mind that the force @xmath55 , see eq .
, depends on how one defines the thickness of the film .
this implies that a quantitative comparison between experimental data and theory is only possible if the data are accompanied by a precise definition of what @xmath0 is .
away from the critical temperature of the system it is customary to write the force acting between the plates of the fluid system in the form @xmath56 where one normally considers the case @xmath57 and omits the apparent dependence on the so - called _ retardation length _ @xcite @xmath58 .
here @xmath59 is the hamaker term , whose dependence from the temperature and chemical potential is given by the so - called hamaker constant @xcite @xmath60 the hamaker constant , as it is clear from above , is a constant only in the sense that it is @xmath0-independent .
it depends on the temperature , chemical potential and on the material properties of the fluid medium and the walls .
the factor @xmath61 in eq .
is introduced there due to historical reason , according to which the interaction energy between two substrates at a finite separation @xmath0 in the case of standard van der waals interaction ( i.e. , @xmath62 ) , away from any phase transition region , is @xcite @xmath63 the hamaker term takes into account the leading @xmath0-dependent parts of the _
i ) _ direct interaction between the slabs @xmath64 , _ ii ) _ between each slab and the fluid medium @xmath65 , and @xmath66 , as well as
the _ iii ) _ interactions between the portion of constituents of the fluid medium situated within the cavity bounded by the substrates @xmath67 , i.e. , @xmath68 note that in eqs .
( [ way ] ) ( [ ham_l ] ) both the slabs and the fluid medium are characterized by their _
properties at the given temperature and chemical potentials .
near the critical temperature @xmath30 of the bulk system eq .
is no longer valid since the critical fluctuations of the order parameter lead to new contribution to the total force called thermodynamic ( critical ) casimir force ( see below ) .
for such a system , following ref .
@xcite , near the bulk critical point eq . can be written in the form @xmath69\nonumber\\ & & + ( \sigma-1)\beta h_a(t,\mu ) l^{-\sigma}\xi_{\mathrm{ret}}^{\sigma - d}.\end{aligned}\ ] ] in eq .
@xmath70 is dimensionless , universal scaling function , @xmath71 and @xmath72 are the temperature and field relevant scaling variables , respectively , while @xmath73 , @xmath74 and @xmath75 are the irrelevant in renormalization group sense scaling variables associated with the interactions in the system . the freedom of choosing the precise definition of what @xmath0 is in systems with boundaries leads to the formal necessity to write @xmath0 as @xmath76 , with @xmath77 being a microscopic length . that will lead to further scaling corrections proportional to @xmath78 .
since in the current article we will keep in all calculated quantities only their leading @xmath0 dependence , we refrain from further refinement of the scaling ansatz . by comparing eq . and
eq . one immediately concludes that @xmath70 tends to zero away from the critical point , i.e. , when at least one of the relevant scaling parameters @xmath79 and @xmath80 becomes large , i.e. , when @xmath81 and/or @xmath82 .
as it is well known , the critical behavior of simple fluids and of binary liquid mixtures is described within the ising universality class which determines the values of the critical exponents @xmath83 and @xmath27 ones the dimensionality of the system @xmath9 is fixed .
when @xmath84 this universality class is characterized by critical exponents @xcite @xmath85 in order to better reflect the actual properties of the non - polar fluids , instead of considering nearest - neighbour interactions we assume long - ranged pair ones between the fluid particles , decaying asymptotically @xmath86 for distances
@xmath87 between each other , and substrate potentials @xmath88 acting on the fluid particles at a distance @xmath89 from each of the two slabs .
we recall that when @xmath90 systems governed by such long - range interactions , usually termed subleading long - ranged interactions @xcite , also belong to the ising universality class characterized by short - ranged forces @xcite .
the last implies , among the others , that the critical exponents , e.g. , do not depend on @xmath91 for such type of interactions .
an important representative of such type of interactions are the non - retarded dispersion interactions with @xmath62 , which are one of the three types of van der walls interactions . by varying the ratio between the strengths of the
long - ranged @xmath92 and the short - ranged
@xmath93 contributions one can quantitatively probe the importance of the long - ranged parts of the interactions within the fluid medium and study potential experiments in colloidal systems which allow for a dedicated tailoring of the form of the effective interactions between colloidal particles . in eq .
, @xmath94 is the standard correction - to - scaling exponent for short - range systems , while @xmath95 and @xmath96 are the correction - to - scaling exponents due to the long - range parts of the interaction potentials between the constituents of the fluid medium and those of the confining walls .
further @xmath97dependent contributions to the total forces @xmath98 such as next - to - leading order contributions to the hamaker terms or higher order corrections to scaling are neglected because they are smaller than those captured in eq . .
the exponent @xmath99 , which appears in the expressions for @xmath100 and @xmath101 , is the standard one characterizing the decay of the bulk two - point correlation function at the critical temperature , @xmath102 is the ( dimensionless ) scaling field associated with the wegner - type corrections , while @xmath103 and @xmath104 are dimensionless non - universal coupling constants : @xmath103 is proportional to the strength @xmath105 of the long - range part of the interaction potential between the particles of the fluid , whereas @xmath104 are proportional to the contrast between the potentials of the bounding slabs and those in the fluid medium ( see below ) . for systems belonging to three - dimensional ising universality class with `` genuine '' non - retarded van der waals interaction one has @xmath62 , and @xmath106 @xcite .
this leads to @xmath107 , @xmath108 , and @xmath109 @xcite . within the mean - field theory with @xmath110 and @xmath111
one has , instead , @xmath112 and @xmath113 .
one then has @xmath114 where all critical exponents take their mean - field values @xmath115 , @xmath116 , @xmath117
. the peculiarities of the scaling theory for systems with dispersion forces are described in refs .
one obtains that despite these systems do belong to the ising universality class with short - ranged forces , the finite - size quantities decay algebraically with @xmath0 towards their bulk values , and not in an exponential in @xmath0 way , when @xmath118 even for @xmath119 , where @xmath120 is the characteristic distance between the molecules of the fluid system . in this regime the dominant finite - size contributions to the free energy and to the force between the walls bounding the system stem from the long - ranged algebraically decaying parts of the interaction potentials .
one can formulate a criterion clarifying when the long - ranged tails of the interactions can not be disregarded even in the critical region of the finite system . neglecting the thickness of the coating layers , for the system under consideration the corresponding criterion states that the long range tails of the interactions can be disregarded only when @xcite @xmath121^{\delta/\nu-\sigma}\ll 1.\end{aligned}\ ] ] using the scaling relations , the above can be formulated as a constraint on the thickness of the system under study .
one obtains that when @xmath6 , where @xmath122^{\nu/\beta},\ ] ] the effects of the long - ranged tails of the interaction can be neglected within the critical region of the finite system . for @xmath62 and with @xmath123 and @xmath27 from eq . for the three - dimensional ising model one obtains @xmath124 we conclude that for moderate values of @xmath0 , i.e. , when @xmath125 the behavior of the casimir force can strongly depend on the details of the fluid - fluid [ the amplitude @xmath126 depends on the details of the fluid - fluid interaction ( see eqs .
( 4.15 ) and ( 4.17 ) in ref .
[ ] and the text therein ) ] and substrate - fluid interactions which , as it turns out , can influence even the _ sign _ of the force . for large @xmath0 , i.e. , for @xmath6 the behavior of the force shall approach the one of the short - ranged system . the last can be easily seen from eq . by simply expanding then the scaling function @xmath70 @xmath127+x_{g}x_{\rm crit}^{g}(x_{t},x_{\mu})\right\}\nonumber\\ & & + ( \sigma-1)\beta h_a(t,\mu ) l^{-\sigma}\xi_{\mathrm{ret}}^{\sigma - d}.\end{aligned}\ ] ] here @xmath128 originates from the short - range interactions
[ see eqs .
( [ xcassranalytic ] ) ] .
it is well known that @xmath129 , i.e. , the force is attractive , under @xmath1 boundary conditions .
@xmath128 provides the _ leading _ behaviour of the force near the bulk critical point ( @xmath130 ) .
there @xmath131 and @xmath132 represent only _ corrections _ to the leading @xmath0 dependance
. the reader can refer for a detailed comment on that matter to ref .
@xcite ( see also ref .
@xcite for the properties of @xmath133 ) . the validity of the proposed criterion as well as the statements made beneath it are well illustrated on fig .
[ fig : short_range_tending ] .
+ the contribution of the dispersion forces to the total effective force @xmath134 can be distinguished from that of the critical casimir force by their temperature dependence , because the leading temperature dependence of the former does not exhibit a singularity .
thus , one has @xmath135 where @xmath136 and @xmath137 one expects that near the bulk critical point @xmath138 where @xmath139 is a scaling function that for large enough @xmath0 ( see below ) approaches the scaling function of the short - ranged system @xmath140 . from eqs .
( [ casimirf_l ] ) , ( [ decomp ] ) ( [ scaling_function ] ) it follows that the scaling function of the critical casimir force @xmath141 is proportional to the sum of @xmath70 and the singular part of the hamaker term .
we will often compare the behavior of the system with subleading long - ranged interactions present with this one of a system with purely short - ranged interactions which will serve as a reference system .
in such a purely short - range system one has @xmath142 . then
, at the bulk critical point @xmath143 the leading term of the thermodynamic casimir force between the slabs bounding the fluid has the form @xmath144 where @xmath145 . here
@xmath146 is an universal dimensionless quantity , called casimir amplitude , which depends on the bulk and surface universality classes ( specified by the boundary conditions @xmath50 ) .
since the casimir force is proportional to @xmath147 the interaction between the walls can become rather strong in a system with high critical temperature such as , e.g. , in classical binary liquid mixtures .
note that the _ sign _ of the force depends on the sign of the casimir amplitude @xmath148 which , on its turn , depends on the boundary conditions @xmath50 . according to the usual convention negative sign corresponds to attraction , while positive sign means repulsion of the surfaces bounding the system .
the experimental and theoretical evidences accumulated till nowadays support the statement that the casimir force is attractive when the boundary conditions on both plates are the same , or similar , and is repulsive when they essentially differ from each other , e.g. , when in the case of a one - component fluid one of the surfaces adsorbs the liquid phase of the fluid while the other prefers the vapor phase .
it is instructive to go back to the explicit physical units in eq . for the physically most relevant case of @xmath84
one has @xmath149 where @xmath150 @xmath151c ( 293.15 k ) .
sine as discussed above , for most systems and boundary conditions @xmath146 , when @xmath152 the thermodynamic casimir force , for some space separation @xmath0 , will be of the same order of magnitude as the quantum one @xmath153 and they both shall be significant and consequently measurable at or below the micrometer length scale .
we turn now to description of the model and the procedure under which our results have been obtained .
we are going to utilize the same type of model already used in refs . @xcite but amended to take into account the specific features of the system considered in the current article . among them
is the role of the two competing substrate potentials . here , in order to introduce the notations needed further , we briefly recall the basic expressions of that model paying a bit more attention only to difference of the current model with that one studied in refs . @xcite .
we consider a lattice - gas model of a fluid confined between two planar walls , separated at a distance @xmath0 from each other , with grand canonical potential @xmath154 $ ] given by @xmath155= k_{b } t \sum_{{\bf r}\in \mathbb{m}}\left\ { \rho({\bf r } ) \ln\left[\rho({\bf r})\right]\right.\nonumber\\ & & \left .
+ \left[1-\rho({\bf r})\right ] \ln\left[1-\rho({\bf r})\right]\right\ } + \dfrac{1}{2}\sum_{{\bf r } , { \bf r}'\in \mathbb{m}}\rho({\bf r})w^{l}({\bf r}-{\bf r}')\rho({\bf r } ' ) \nonumber \\ & & + \sum_{{\bf r } \in \mathbb{m}}\left[v^{(s_{1}|l|s_{2})}(z)-\mu\right ] \rho({\bf r}),\label{garand_pot_l_lattice_model}\end{aligned}\ ] ] where @xmath34 is a simple cubic lattice in the region occupied by the fluid medium @xmath156 $ ] and @xmath157 is an external potential that reflects the interactions between the confining walls and the constituents of the fluid , given by @xmath158 where @xmath159 , with @xmath160 this type of functional can be viewed as a modification of the model utilized by fisher and nakanishi @xcite in their mean - field investigation of systems governed by short - range forces . in eq .
the terms in curly brackets multiplied by @xmath48 correspond to the entropy contributions to the total energy , @xmath161 is the non - local coupling ( interaction potential ) between the constituents of the confined medium and @xmath4 is the chemical potential . here and in the remainder of this paper , all length scales are taken in units of the lattice constant @xmath120 ( for concrete values see table .
[ table_fluid_prop ] ) , so that the particle number density @xmath162 becomes simply a number density which varies in the range @xmath163 $ ] .
the variation of eq . with respect to @xmath162 leads to the equation of state for the equilibrium density @xmath164 @xmath165 \right\}.\label{eqs_of_state_l } \end{aligned}\ ] ] the advantage of this type of equation is that it lends itself to numerical solution by iterative procedures . for a given geometry and external walls
- fluid potential @xmath157 its solution determines the equilibrium order parameter profile @xmath166 in the system . inserting this profile into eq . , renders the grand canonical potential of the considered system . denoting , as in refs .
@xcite @xmath167 and @xmath168 , where @xmath169 , the equation of state eq .
can be rewritten in the standard form @xmath170\right\}.\end{aligned}\ ] ] the bulk properties of the model are well known ( see , e.g. @xcite and the references therein ) .
we recall that the order parameter @xmath171 of the system has a critical value @xmath172 which corresponds to @xmath173 so that @xmath174 .
the bulk critical point of the model is given by @xmath175^{-1 } , \mu=\mu_{c}=-2\sum_{\bf r}j^{l}({\bf r})\}$ ] with the sum running over the whole lattice .
within the mean - field approximation the critical exponents for the order parameter and the compressibility are @xmath176 and @xmath177 , respectively .
the _ effective _ surface potential @xmath178 in eq .
is given by @xmath179 where @xmath180 , and contributions of the order of @xmath181 , @xmath182 , etc .
, have been neglected , @xmath183 are ( @xmath3- and @xmath4-independent ) constants , @xmath184 is a proper lattice version of @xmath185 as the interaction energy between the fluid particles , and @xmath186 is the one between a fluid particle and a substrate particle , @xmath187 is the discrete delta function and @xmath188 is the heaviside step function with the convention @xmath189 ; in eq .
@xmath190 are the number densities of the coated slabs in units of @xmath191 ( for concrete values see table .
[ table_fluid_sub ] ) .
note that the effective potentials @xmath192 result from the difference between the relative strength of the substrate - fluid interactions for substrates with density @xmath190 and that of the fluid - fluid interactions for a fluid with density @xmath193 .
the restriction @xmath194 holds because we consider the layers closest to the substrate to be completely occupied by the liquid phase of the fluid ( which implies that we consider the strong adsorption limit ) , i.e. , @xmath195 , which is achieved by taking the limits @xmath196 ; thus , the actual values of @xmath197 will play no role . in order to preserve the monotonic behavior of @xmath198 as a function of the distance @xmath87 between the particles , in eq .
we have to require that @xmath199 , i.e. , @xmath200 , where @xmath201 from eqs .
( [ eq_state_standard_l_ab ] ) ( [ jlsdeftext ] ) one can identify the dimensionless coupling constants @xmath202 , @xmath203 appearing in eq .
@xmath204 here @xmath205 , i.e. , @xmath206 corresponds to walls `` preferring '' the liquid phase of the fluid , while @xmath207 , or @xmath208 mirrors the one with affinity to its gas phase .
the marginal case @xmath209 which corresponds to @xmath210 will be commented further in the text . in terms of @xmath211 the functional defined by eq
. takes the form @xmath212=k_{b}t\sum_{{\bf r}\in \mathbb{m } } \left\{\dfrac{1+\phi({\bf r})}{2 } \ln\left[\dfrac{1+\phi({\bf r})}{2}\right]\right .
\nonumber \\ & & \left.+ \dfrac{1-\phi({\bf r})}{2}\ln\left[\dfrac{1-\phi({\bf r})}{2}\right]\right\ } -\dfrac{1}{2}\sum_{{\bf r } \in \mathbb{m}}[\delta\mu-\delta v(z ) ] \phi({\bf r } ) \nonumber\\ & & -\dfrac{1}{2}\sum_{{\bf r } , { \bf r}'\in \mathbb{m}}j^{l}\left({\bf r}-{\bf r}'\right)\phi({\bf r})\phi\left({\bf r}'\right)+\omega_{\rm reg},\ \ \ \ \ \ \\end{aligned}\ ] ] where @xmath213\label{omega_reg_l}\end{aligned}\ ] ] do not depend on @xmath214 and , therefore , is a regular background term carrying an @xmath97dependence and thus showing up in the corresponding force . the only quantity in eq . , which still has to be identified for our model , is the value of the coupling constant @xmath103 .
according to ref .
@xcite @xmath215 where @xmath216 and @xmath217 are coefficients in the fourier transform @xmath218 of the interaction @xmath219 [ see eq . ]
( for more details see ref .
it turns out that @xmath103 depends on @xmath220 . in accordance with eq .
, for the finite - size behavior of the excess grand canonical potential per unit area @xmath47 of a liquid film in the case when both confining surfaces strongly adsorb the liquid phase , one expects @xmath221 where @xmath222 and @xmath223 are the non - singular parts of the surface tensions at the surfaces of the confining walls , while the singular parts are incorporated in the scaling function @xmath224 , which arguments are as those of the function @xmath70 in eq . . in eq .
@xmath59 is the hamaker term which explicit form is derived in appendix [ sec : hamakerterm ] .
the result is @xmath225 note that this result is in full agreement with the dzyaloshinskii - lifshitz - pitaevskii ( dlp ) theory @xcite .
it provides , however , an easy possibility to study the @xmath3 and @xmath4 dependence of @xmath226 by studying the corresponding dependencies of @xmath227 , say , near the critical point of the fluid system . in accord with the dlp theory it teaches us that the _ sign _ of the hamaker term depends on the _ contrast _ of material properties of the two bounding substrates with respect to the substance which is in - between them .
coating the substrate surfaces of the system with some additional material does not change the leading - order @xmath0 dependence of the interactions between the substrates and , therefore , does not change the above property . let us stress that when @xmath228 , i.e. , when the bounding substances are made from same material , the hamaker term will be _ negative _ , independent on the properties of the fluid in - between them , i.e. , @xmath229 which corresponds to _ attraction _ between the confining walls .
similar is the situation when @xmath230 , i.e. , when the fluid separating the substrates is replaced by vacuum .
then , again , @xmath231 _ irrespective _ of the material properties of the bounding substances , since @xmath232 . further details on the behavior of @xmath59 are given in appendix [ sec : hamakerterm ] . in the next section [ sec : finsizebehav ]
we are going to present results for @xmath233 based on the model described in the current section .
let us start by rewriting the equations presented in sec .
[ sec : model ] in the form suitable for studying of a system in a film geometry .
because of the translational symmetry of the system along the bounding surfaces , the quantities of interest depend only on the spacial coordinate @xmath89 along which the system is finite . thus , one can write : @xmath234 , where @xmath235 , i.e. , the local order parameter profile is given by @xmath236 , with @xmath237 .
hence the equation eq
. for the equilibrium profile becomes @xmath238=\dfrac{\beta}{2}\left[\delta\mu-\delta v(z)\right]\nonumber\\ & & + k\left\{\text{\ { a}}_{d,\sigma}\left(\lambda\right)\phi^{*}(z)+\text{\ { a}}_{d,\sigma}^{nn}\left(\lambda\right)\left[\phi^{*}(z+1)+\phi^{*}(z-1)\right]\right.\nonumber\\&&\left.+\lambda \sum_{z'=0}^{l}g_{d,\sigma}(|z - z'|)\theta(|z - z'|-1)\phi^{*}(z')\right\},\end{aligned}\ ] ] where @xmath168 , @xmath178 is defined in eq . , @xmath239 , @xmath240 , @xmath241 with @xmath242 .
the functions @xmath243 , @xmath244 and @xmath245 are determined in eqs .
( c10 ) , ( c11 ) and ( c12 ) of ref .
@xcite , respectively .
when @xmath246 the equilibrium finite - size order parameter @xmath247 tends to its bulk value @xmath248 $ ] . the corresponding equation for @xmath249 , performing the limit @xmath250 in eq . , reads @xmath251\phi_{b}\right\},\end{aligned}\ ] ] wherefrom one immediately identifies the coordinates of bulk critical point @xmath252^{-1},\ \ \ \ \\end{aligned}\ ] ] and @xmath253 .
note that the position of the bulk critical point depends on @xmath220 , i.e. , on the presence and the strength of the long - ranged tails in the fluid - fluid interactions .
for a fluid confined to a film geometry the natural quantity to consider is the excess grand canonical potential per unit area @xmath47 : @xmath254-l\omega_{\rm bulk}$ ] . using the result of ref . @xcite
( 3.14 ) there ] , as well as the identity @xmath255 $ ] , one can write @xmath256 in the form @xmath257-\dfrac{1}{2}\ln\left[1-\phi_{b}^{2}\right]\right.\nonumber\\&&\left . + \dfrac{1}{4}\phi^{*}(z)\ln\left[\dfrac{1+\phi^{*}(z)}{1-\phi^{*}(z)}\right]-\dfrac{1}{4}\phi_{b}\ln\left[\dfrac{1+\phi_{b}}{1-\phi_{b}}\right]\right .
\nonumber\\&&\left.+\dfrac{1}{2}\beta\delta v(z)\phi^{*}(z)-\dfrac{\beta\delta\mu}{2}[\phi^{*}(z)-\phi_{b}]\right\}+\beta\omega_{\rm reg},\ \ \ \ \\end{aligned}\ ] ] where @xmath258\nonumber\\ & & \times l^{-\sigma+1}\xi_{\rm ret}^{\sigma - d}.\end{aligned}\ ] ] as it is clear from eq . ,
the model presented above does not account for the direct wall - wall interaction , but only for the walls - fluid and fluid - fluid ones .
thus , in order to obtain the complete net force acting between the surfaces bounding the fluid one has to add to the force calculated from eq . via eq .
the one due to the direct wall - wall interaction .
then the resulting net force will be the _ total _
force @xmath259 between the plates bounding the fluid . with the use of eqs .
( [ def ] ) , ( [ bethasigmaminusonehal ] ) and ( [ deltaomegaexz_equilibrium ] ) one can write @xmath259 in the form @xmath260\nonumber\\ & & -\dfrac{4ks_{1,c}s_{2,c}}{g(d,\sigma)k_{c}^{2}\lambda}l^{-\sigma}\xi_{\rm ret}^{\sigma - d},\end{aligned}\ ] ] where the last term represents the direct wall - wall interaction .
on the other hand , if one subtracts from the potential @xmath261 its regular part @xmath262 , i.e. , if we consider the quantity @xmath263-l\omega_{\rm bulk},\ ] ] then , in accord with eqs .
( [ decomp ] ) ( [ def_casimir ] ) , the @xmath0 dependence of @xmath264 via eq .
provides the singular part of the total force , i.e. , the _
critical _ casimir force @xmath265 .
explicitly , one has @xmath266.\end{aligned}\ ] ] obviously , near @xmath30 the total and the casimir force are related via the expression @xmath267l^{-\sigma}\xi_{\rm ret}^{\sigma - d}.\end{aligned}\ ] ] eqs .
( [ totalforcemeanfieldgeneral ] ) , ( [ order_parameter_equation_d4 ] ) ( [ totcasrelation_new ] ) provide the basis for our numerical treatment of the finite - size behavior of the total as well as of the casimir force in fluid systems governed by van der waals interactions .
let us briefly outline the main steps of this numerical procedure .
we start by determining the equilibrium order - parameter profile [ @xmath268 , solving iteratively eq . .
the solution of this equation depends , however , for a given range of parameters @xmath3 and @xmath269 , on the choice of the initial state of the order - parameter profile .
the two basic initial states are i : : a `` liquid - like '' state in which all the sites of the lattice are occupied by a fluid particle , i.e. [ @xmath270 ; ii : : a `` gas - like '' state where @xmath271 . thus one needs to calculate the profile starting from both initial states . if the two final states coincide they provide the unique minimum of the functional see eq . .
if they differ , one has to check which one of them provides the absolute minimum of the grand canonical potential .
the simplest way to clarify that question is to calculate @xmath272 via eq . .
on a given line in the @xmath273 plane the values of the grand canonical potential of the two profiles coincide .
the set of points on this line defines the phase diagram of the system .
an alternative and practically more effective approach for determining of the phase diagram will be presented in sec .
[ sec : phasebehaviour ] . heretofore for systems governed by dispersion interactions
there are no closed - form analytical expressions about the scaling function of the casimir force even within mean - filed theory .
such expressions exist , however , for systems governed by short - range interactions @xcite . for @xmath1 boundary conditions at @xmath253 the corresponding results are [ xcassranalytic ] @xmath274^{4}k^{2}\left(1-k^{2}\right),\ ] ] with @xmath275^{2}\left(2k^{2}-1\right)$ ] ; @xmath276 with @xmath275^{2}\left(2k^{2}-1\right)$ ] ; @xmath277 with @xmath278^{2}\left(k^{2}+1\right)$ ] , where @xmath279 is the complete elliptic integral of the first kind , @xmath280 . in eq .
the scaling variable of @xmath281 is @xmath282 .
we note that @xmath281 is _ analytic _ for _ all _ values of @xmath283 , because the film critical point @xmath284 $ ] is located off coexistence at @xmath285 @xcite .
obviously one has @xmath286 if @xmath287 ( with @xmath288 corresponding to the bulk critical point ) , @xmath289 if @xmath290 ( with @xmath291 corresponding to the actual film critical point ) , and @xmath292 if @xmath293 ( negative @xmath294 describes the region below the bulk critical point ) .
for the field dependent scaling function of the casimir force @xmath295 there are _ no _ analytical expressions even for short - range systems .
+ + + in the next section , based on the results reported in sections [ sec : model ] [ sec : finsizebehav ] , we present numerical results for the behaviour of the scaling functions of the critical casimir force and of the van der waals term for the cases @xmath110 and @xmath62 .
in the current section , using the results for @xmath110 from the mean - field type numerical study of the system introduced in sections [ sec : model ] [ sec : finsizebehav ] , we will present some approximate results for the behavior @xmath70 , the hamaker term and the resulting critical casimir force , between the two surfaces confining a van der waals type fluid in @xmath84 . within the mean - field theory the dependance of the corresponding forces from the temperature @xmath3 and the chemical potential difference @xmath269 near the bulk critical point @xmath297
is given through the dimensionless temperature and field scaling variables @xmath298 and @xmath299 . in our numerical treatment
we take these variables to range in the intervals : @xmath300\equiv[-576;576]$ ] and @xmath301\equiv[-13824;13824]$ ] .
we consider systems with fluid layers @xmath0 being @xmath302 .
we fix one of the walls - fluid coupling parameters ( say @xmath303 ) to have a value either @xmath304 or @xmath305 , while the other one , @xmath306 , is varied from @xmath307 to @xmath305 with a step of @xmath308 .
the fluid - fluid coupling parameter @xmath220 is supposed to be either @xmath304 or @xmath309 . in order to determine the behaviour of the above mentioned quantities in a @xmath84 confined fluid system , using the presented approximation
, we proceed in the following way .
first , within the mean - field treatment of the critical behaviour @xmath310 we solve the equation for the equilibrium order parameter profile eq . .
next , with the use of eqs .
( [ deltaomegaexz_equilibrium ] ) and ( [ totalforcegeneralexpression_new ] ) we obtain the total force of interaction between the plates . subtracting from eq .
the hamaker term eq .
we end up with the scaling function @xmath70 [ see also eq . ] . in order
that this function contributes properly to the critical casimir and hence to the total force of interaction in @xmath311 , one must normalize it accordingly .
the need of such normalization is explained in details in ref .
@xcite ( see there sections iv.a.1 and iv.a.3 ) . for boundary conditions
@xmath50 one has @xmath312 ^ 4 x^{(\tau),\mathrm{mf}}_{\rm crit}(\cdot ) , \ \ \ \ \ \\end{aligned}\ ] ] where @xmath313 is the value of the scaling function for a system within mean - field treatment governed by short - range interactions at its corresponding bulk critical point , and @xmath314 is the scaling function of the critical casimir force , calculated for @xmath110 , with @xmath315 .
here @xmath316 is the casimir amplitude for the @xmath84 ising universality class with boundary conditions @xmath50 , while @xmath317 and @xmath318 are the amplitudes of the bulk correlation length in mean - field systems with , correspondingly , short - ranged ( @xmath319 ) and long - ranged ( @xmath320 ) fluid - fluid interactions , as @xmath321 ( see eqs .
( 4.15 ) and ( 4.17 ) in ref .
[ ] ) . therefore for @xmath322 one has @xmath323 , and hence @xmath324 respectively .
taking into account that the value of the casimir amplitude for the @xmath84 ising universality class with @xmath1 boundary conditions @xcite is @xmath325 and that @xmath326 , for the normalizing coefficients one obtains : @xmath327 for @xmath328 and @xmath329 when @xmath330 .
the dependence of the scaling function @xmath70 on the temperature @xmath283 and field @xmath331 scaling variables is summarized in figs .
[ fig : cfspsm ] and [ fig : cfsmsm ] for different separations @xmath0 .
once having a good approximation of @xmath70 in @xmath84 , obtained in the way described above , one can determine the scaling function of the critical casimir force of the fluid system by adding to @xmath70 the singular part of the hamaker term given by eq .
, i.e. , that part of it that depends on @xmath332 .
when performing this procedure one must pay attention to the use of the coupling parameters @xmath333 , because they also depend on the dimensionality @xmath9 of the system .
using eqs .
( [ v_s_l_def ] ) , ( [ deltav_s_l_def ] ) and ( [ s_def_l ] ) one has @xmath334 as it is clear from the above , in system governed by short - range interactions @xmath70 coincides with @xmath335 ; @xmath335 is _ negative _ , which corresponds to _ attractive _ critical casimir force , for any value of the scaling variables @xmath336 and @xmath337 under @xmath1 boundary conditions see eq . for @xmath335 in the case @xmath338 , as well as fig .
[ fig : short_range_tending]@xmath339 and @xmath340 . note that under such boundary conditions one observes an enhancement of the order parameter due to the confinement in comparison to the order parameter at the same distance from an individual wall . on that ground in systems with long - ranged interactions such that @xmath341 and @xmath342 we expect @xmath70 to remain again negative for any separation @xmath0 and at any value of the scaling variables , irrespective of the value of @xmath220 ( which is always non - negative ) .
indeed , this turns out to be true and is depicted in fig .
[ fig : cfspsm](*a * ) and ( * b * ) . however when @xmath343 and the separation between the walls is relatively small a significant part of the system is disordered which results in non - negative or sign - changing scaling function [ see fig .
[ fig : cfspsm](*c * ) and ( * d * ) ] . as the distance @xmath0
is increased the influence of the effective surface potential @xmath178 [ see eq . ] quickly decreases and only the additional ordering effect of the fluid - fluid interactions influences the behavior of the order parameter and hence of @xmath70 . along with the same line of arguing , when both wall - fluid coupling parameters are negative
, @xmath70 is negative for any @xmath283 and @xmath331 only for very large separations @xmath0 where the effect of the long - ranged interactions on the behavior of the system is negligible .
naturally , since the short - ranged surface potentials do support @xmath1 boundary conditions , the role of the negative substrate potentials , which oppose the order near the boundary , will be stronger than that of the positive substrate potentials which try to reinforce the phase preferred near the boundary .
for example , we observe that the behavior of @xmath70 in a system with @xmath344 and in such with @xmath345 is almost identical for any @xmath0 , both as a function of @xmath283 and @xmath331 [ compare fig .
[ fig : cfspsm](*c * ) and ( * d * ) and fig . [
fig : cfsmsm](*a * ) and ( * b * ) ] @xcite . thus , for a fixed @xmath220 the behavior of the scaling function
is mainly determined by the interplay between the short - ranged surface fields and the strong negative wall - fluid coupling @xmath306 .
if @xmath346 and @xmath347 the scaling function @xmath70 exhibits an unexpected behavior as a function of @xmath0 : for moderate values of @xmath0 the maximum of the _ repulsive _ part of the force increases with increasing @xmath0 both as a function of @xmath283 and @xmath331 [ see fig .
[ fig : cfsmsm](*c * ) and ( * d * ) ] ; for larger values of @xmath0 the maximum decreases , as expected , and the overall behavior of the scaling function approaches that one of the system with completely short - ranged interactions .
the behavior of the critical casimir force @xmath141 is depicted on fig .
[ fig : forceinterplay ] . in order to illustrate only the main idea , here we restrict ourself to the choice of parameters @xmath348 , @xmath349 while @xmath350 .
the data for the casimir force and the hamaker term are presented for a system with @xmath351 layers .
first , let us recall that for @xmath352 at @xmath253 one has @xmath353 and , hence , the behavior of @xmath141 and that of @xmath70 coincide in this region [ see fig . [
fig : forceinterplay](*a * ) and ( * c * ) ] . on the other hand ,
if @xmath354 one has @xmath355 , with the singular ( i.e. , the @xmath249 dependent ) part of the hamaker term corresponding to repulsion for @xmath341 and attraction otherwise , see eq . .
for @xmath356 the scaling function @xmath70 is negative , but the singular part of the hamaker term is positive , i.e. , repulsive in the low - temperature region .
thus , the casimir force is attractive near and above @xmath30 but becomes repulsive below @xmath30 . for @xmath357 one
has that @xmath70 is non - negative , but with the singular part of the hamaker term being negative , i.e. , attractive for @xmath358 .
thus , the resulting critical casimir force @xmath141 changes sign from being repulsive above @xmath30 to becoming attractive slightly below @xmath30 .
the behaviour of the casimir force as a function of @xmath331 at @xmath359 for the two sub - cases of @xmath350 is depicted on fig .
[ fig : forceinterplay](*b * ) and ( * d * ) . for @xmath356
one observes that @xmath360 , while the singular part of the hamaker term changes sign from positive to negative with @xmath361 decreasing .
thus , the resulting casimir force changes sign once from being slightly positive ( repulsive ) for @xmath362 to negative ( attractive ) for @xmath363 .
the case @xmath357 is much more interesting .
since then @xmath70 changes sign twice for @xmath364 , while the singular part of the hamaker term changes sign ones , the resulting casimir force happens to change its sign tree times as a function of @xmath337 .
one might pose the question about the phase behavior of the thin fluid systems with such competing surface and substrate potentials .
we consider that question in the next section .
confined finite - size fluid system with strong adsorption of the `` liquid '' phase of the fluid medium on the confining walls , for different separations @xmath0 between them .
the values of the coupling parameters characterizing the interactions in the system are @xmath365 , and the coordinates of the finite - size critical points for the different separations are : bottom - up @xmath351 ( @xmath366 ) @xmath367 ; @xmath368 ( @xmath369 ) @xmath370 ; @xmath371 ( @xmath372 ) @xmath373 . ]
the phase behaviour of a confined fluid medium between parallel walls exerting identical surface adsorption potentials @xmath374 on both surface layers has been studied extensively both theoretically and experimentally @xcite .
so far the considerations have been made either for fluid systems governed by pure short - range interactions @xcite , or for systems with surface potentials strongly preferring the same phase of the fluid @xcite .
here we are going to consider the case when at least one of the wall - fluid potentials favor a phase of the fluid different from the one preferred by the short - ranged surface potentials .
the phase behavior of systems with @xmath375 and of the bulk system with @xmath356 , @xmath376 and @xmath348 , obtained within our mean - field model , is presented in fig .
[ fig : pddiffl ] .
a relatively detailed study of the influence of the different values of @xmath303 , @xmath306 and @xmath220 on the phase behavior of thin fluid films ( in the case considered @xmath351 ) is illustrated in figs .
[ fig : pddiffsandlambda ] and [ fig : ncpd ] , where fig .
[ fig : ncpd ] represents blow - up views of some of the phase diagrams very close to the capillary condensation critical points @xmath377 .
confined finite - size fluid systems with strong adsorption of the `` liquid '' phase of the fluid medium on the confining walls , obtained for @xmath351 and various values of the coupling parameters : * ( a ) * bottom - up ( ) @xmath378 ; ( ) @xmath379 ; ( ) @xmath380 ; ( ) @xmath381 ; ( ) @xmath344 ; * ( b ) * bottom - up ( ) @xmath380 ; ( ) @xmath382 ; ( ) @xmath383 ; ( ) @xmath384 ; ( ) @xmath385 .
the coordinates of the finite - size critical points @xmath386 $ ] of the different systems are : * ( a ) * bottom - up ( @xmath387 ) @xmath388 ; ( @xmath389 ) @xmath390 ; ( @xmath366 ) @xmath367 ; ( ) @xmath391 ; ( @xmath392 ) @xmath393 ; * ( b ) * bottom - up ( @xmath394 ) @xmath395 ; ( ) @xmath391 ; ( ) @xmath396 ; ( ) @xmath397 ; ( ) @xmath398 . the phase diagram of a bulk system @xmath399 is shown as thick black line ending at the bulk critical point ( ) .,title="fig : " ] confined finite - size fluid systems with strong adsorption of the `` liquid '' phase of the fluid medium on the confining walls , obtained for @xmath351 and various values of the coupling parameters : * ( a ) * bottom - up ( ) @xmath378 ; ( ) @xmath379 ; ( ) @xmath380 ; ( ) @xmath381 ; ( ) @xmath344 ; * ( b ) * bottom - up ( ) @xmath380 ; ( ) @xmath382 ; ( ) @xmath383 ; ( ) @xmath384 ; ( ) @xmath385 .
the coordinates of the finite - size critical points @xmath386 $ ] of the different systems are : * ( a ) * bottom - up ( @xmath387 ) @xmath388 ; ( @xmath389 ) @xmath390 ; ( @xmath366 ) @xmath367 ; ( ) @xmath391 ; ( @xmath392 ) @xmath393 ; * ( b ) * bottom - up ( @xmath394 ) @xmath395 ; ( ) @xmath391 ; ( ) @xmath396 ; ( ) @xmath397 ; ( ) @xmath398 . the phase diagram of a bulk system @xmath399 is shown as thick black line ending at the bulk critical point ( ) .,title="fig : " ] [ ( ) , ( ) and ( ) ] and the phase diagram ( ) of a @xmath400 finite - size fluid system , confined between parallel walls both strongly adsorbing the `` liquid '' phase of the fluid medium , for @xmath351 and values of the coupling parameters @xmath401 .
the bulk critical point is denoted by the symbol , while the finite - size critical point @xmath402 is marked by the symbol @xmath389 .
the curve designated by ( ) shows the behavior of @xmath403 at @xmath404 .
this behavior at the finite - size critical point @xmath402 is depicted with the symbol ( ) .
when @xmath405 and @xmath406 , the compressibility @xmath407 changes with a finite jump ( ) , being an indicator of a first - order phase transition .
the points of first - order phase transition form the coexistence curve , i.e. , determine the phase diagram in the phase plane . ]
+ we observe that the overall phase behavior of the system is similar to that one obtained for a short - ranged system by fisher and nakanishi @xcite one has a line of a capillary condensation shifted towards the gas bulk phase which terminates at its own finite - size critical point @xmath377 .
one can identify the following properties of the phase diagrams presented in figs .
[ fig : pddiffl ] , [ fig : pddiffsandlambda ] and [ fig : ncpd ] : _ i ) _ the precise position of the phase diagram depends on @xmath0 .
this property is well known also for systems with short - ranged interaction .
_ ii ) _ for a fixed @xmath0 the precise position of the phase diagram depends on @xmath303 , @xmath306 and @xmath220 smaller @xmath0 stronger the corresponding influence of these parameters .
negative wall - fluid couplings @xmath303 and/or @xmath306 have disordering effect on the system which manifests in increase of the gas phase and decrease of the liquid one near the surfaces .
generally the phase diagram , as then expected , appears above the one of a short - range systems [ see fig .
[ fig : pddiffsandlambda](*b * ) ] since in order to `` liquify '' the system one now needs higher `` pressure '' .
_ iii ) _ the position of @xmath377 , thus the shift of the critical point of the finite system with respect to that one of the bulk one , depends not only on @xmath0 , as in the short - ranged systems , but also on @xmath303 , @xmath306 and @xmath220 . for @xmath408 the ordering in the system increases and , hence , one observes increase in both the critical temperature shifts as well as in the chemical potential difference compared to systems governed by pure short - range interactions . when @xmath303 and @xmath306 are both positive this effect is even further enhanced .
if , on the other hand , for fixed @xmath409 the parameters @xmath303 and @xmath306 are both negative or , at least , have different signs , the shift of the finite - size critical point with respect to that one in short - range systems can be completely canceled or , for strong enough potentials , this shift can even further decrease via @xmath377 moving toward the corresponding bulk critical point . _
iv ) _ the competing effect of @xmath3 , @xmath269 , @xmath303 , @xmath306 and @xmath220 on the ordering in the system leads near the capillary condensation critical point for thin films to a non - monotonic behavior of the phase diagram
[ fig : ncpd ] .
the behavior of the system with at least one of the wall - fluid potentials being negative is especially interesting see fig .
[ fig : ncpd](*d * ) , where in a small temperature interval at phase coexistence when @xmath3 decreases @xmath361 increases .
below we briefly explain how the above phase diagrams have been calculated . on fig .
[ fig : pdsusceptibility ] we illustrate the method used to obtain the phase diagram and the finite - size critical point at which the coexistence curve ends of finite systems characterized by long - ranged fluid - wall potentials .
we do that by studying within our mean - field model the scaling functions of the total isothermal compressibility @xmath403 .
when @xmath404 one has that @xmath403 is a smooth function of the temperature , even in the vicinity of the bulk critical temperature , reaching its maximum for @xmath352 .
for @xmath410 the value of the maximum increases , being still finite , and appearing at @xmath354 . upon reaching the finite - size critical point @xmath402
, @xmath403 diverges , giving rise to power - law singularity . at temperatures @xmath405 and chemical potentials @xmath406 , the value of @xmath403 changes with a finite jump , being an indicator of a first - order phase transition .
the points of first - order phase transition form a line the coexistence curve in the phase plane , separating the observed phases : spin-``up '' phase in magnetic systems , liquid phase in one - component fluid systems , a - rich phase in binary liquid mixtures and , correspondingly , spin-``down '' phase , gas phase , b - rich phase . at the critical point and beyond
the differences between these phases disappear .
the resulting phase diagrams are shown in fig .
[ fig : pddiffsandlambda ] , where ( * a * ) summarizes the curves with @xmath411 , and ( * b * ) those with @xmath412 , where different combinations of values of @xmath413 and @xmath220 are considered .
the line marked with filled ( red ) circles represents the phase diagram for a system with completely short - ranged interaction . as it was shown in refs .
@xcite , for the total isothermal compressibility per particle ( or susceptibility , if one considers magnetic systems ) , one has @xmath414 where @xmath415 is the inverse of the matrix @xmath416 with elements @xmath417 and @xmath418,\end{aligned}\ ] ] is the `` local '' isothermal compressibility , which reflects the response of the system from a given layer due to the change of the external field in that layer .
more precisely , in eq .
@xmath419 is the density - density correlation function , @xmath420 $ ] , where the functional derivative is taken with respect to the field @xmath421/2 $ ] . in order to determine @xmath422 or its `` scaling function '' @xmath423 in a fluid film with thickness @xmath0
, we first solve numerically eq . within our mean - field model , which allows us to determine the matrix @xmath416 with the use of eq .
after that , using eq .
, we obtain the `` local '' isothermal compressibility @xmath424 , and summing over @xmath89 , the total isothermal compressibility .
we recall that within the mean - field treatment one has @xmath425 and @xmath116 .
the phase diagrams depicted on fig .
[ fig : pddiffsandlambda ] result from the interplay between the fluid - fluid and walls - fluid interactions .
note that the increase of @xmath220 at fixed @xmath333 lowers @xmath426 at coexistence , while the decrease of the wall - fluid coupling at one of the walls ( say @xmath306 ) keeping @xmath220 and @xmath303 constants , increases it .
in the fast emerging field of micro- and nano- devices , the fluctuation induced and dispersion forces like the casimir and van der waals ones , associated with the interaction between the individual components of such devices , play an essential role . due to the quantum origin of these forces , currently there are no theoretical as well as experimentally suggested control parameters that can be used to conveniently and reversibly govern the sign and the strength of these forces if the interacting objects are in vacuum .
the situation changes if these interactions take place in a fluid . very recently in ref .
@xcite the authors suggested controlled quantum casimir levitation caused by the introduction of thin film surface coating on a porous planar substrates both immersed in a suitable fluid .
more specifically , they have shown that the casimir force between teflon and cassiterite ( @xmath427 ) nanosheet immersed in cyclodecane is attractive at large separations but repulsive at small separations , resulting in a stable equilibrium distance where the total force is zero . in ref .
@xcite the authors reported the use of aerogels , yielding repulsion down to submicron distances at realistic porosities .
the described so far set - ups are quite close in spirit to the considerations presented in the current article with the basic difference that the force will be due to the critical fluctuations of the fluid , instead of the fluctuation of the electromagnetic field .
we suppose that the usage of low density substrates like aerogels can be also applied in the thermodynamic casimir effect since the parameters @xmath428 directly depend , as shown in eqs .
( [ deltav_s_l_def ] ) and ( [ s_def_l ] ) , on the densities of the slabs .
.[table_fluid_prop ] physical characteristics of the considered confined fluids ( column 1 ) the critical density @xmath193 in units @xmath429 ( column 2 ) , temperature @xmath430 measured in @xmath431 ( column 3 ) , pressure @xmath432 in @xmath433 ( column 4 ) , the lattice constant @xmath120 in ( column 5 ) , the distance @xmath434 also in ( column 6 ) at which the inter - particle potential is zero and the potential well depth @xmath435 in units @xmath147 ( column 7 ) . the values of @xmath436 and @xmath432 for argon are taken from ref . [ ] , for krypton from ref . [ ] and for xenon from ref . [ ] . the value of @xmath120 is calculated based on the data for @xmath193 ( see the text ) and
the presented values of @xmath434 and @xmath435 are taken from ref .
[ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] in order to show with concrete examples that the experimental observation of the effects theoretically predicted in the current article is possible , we consider the occurrence of critical fluctuations in simple non - polar fluids such as argon ( ar ) , krypton ( kr ) or xenon ( xe ) , confined between graphene coated carbon nanotube aerogel substrate ( cas ) @xcite and a silver ( ag ) or gold ( au ) one .
both confining surfaces are assumed coated by monatomic ( or very thin ) lead ( pb ) film , ensuring the ( + , + ) boundary conditions . the basic data needed for the calculations concerning the considered fluids are presented in table [ table_fluid_prop ] .
furthermore , from the data reported in ref .
[ ] we have that @xmath437 and @xmath438 , where @xmath439 , see table [ table_fluid_prop ] , is specific for each fluid medium . as far as the geometry of the experimental set - up matters , one can choose to study the predicted behavior of the force either in plane parallel geometry @xcite , or consider such where one of the plates is flat and the other has the shape of a spherical lens with large @xmath440 curvature radius @xcite . as pointed out in ref .
[ ] , the difficulties in the usage of the plane parallel geometry are bigger , which leads to lower accuracy of the results in comparison to the other configuration
. the use of plates in the form of spherical lens to study effects predicted for systems composed out of flat parallel surfaces can be mathematically justified , as usual , by the derjaguin @xcite or the gradient expansion @xcite approaches , which are applicable for calculations of the interactions between curved objects . in what follows we evaluate the coupling parameters @xmath441 and @xmath220 which are the basic parameters reflecting the material properties in the theory for the behavior of the force presented above
[ fig : cfspsm ] [ fig : forceinterplay ] . from eqs .
( [ deltav_s_l_def ] ) and ( [ s_def_l ] ) we have that @xmath442 $ ] , and thus , in order to determine @xmath441 , one needs to know the long - range inter - particle interaction energies @xmath443 and @xmath92 , as well as the number density @xmath444 ( see column 7 in table [ table_fluid_sub ] ) of the confining substrate relative to that of the fluid medium @xmath193 .
since within the mean - field theory the number density of the fluid is @xmath173 one has @xmath445/\rho_{c}[\rm mol / cm^{3}]$ ] , with @xmath446=\rho[{\rm g / cm^{3}}]/(a_{{\rm r}}u n_{a})$ ] , where @xmath447 is the standard atomic weight of the substance , @xmath448 the atomic mass unit and @xmath449 is the avogadro constant . from here
one can also estimate the lattice constant of the fluid medium at the critical point ; one has @xmath450}=10^{10}\times\{\rho[{\rm g / cm^{3}}]/(a_{{\rm r}}u)\}^{-1/3}$ ] ( see column 5 in table [ table_fluid_sub ] ) .
we will make the assumption that all interactions between the constituents of the system are of lennard - jones type , i.e. , the interaction potential can be written in the form @xmath451\nonumber\\ & & = j_{\rm sr}\left[\left(\dfrac{r_{m}}{r}\right)^{12}-2\left(\dfrac{r_{m}}{r}\right)^{6}\right],\end{aligned}\ ] ] where @xmath452 is the distance at which the potential reaches its minimum , @xmath453 is such that @xmath454 and @xmath455 is the depth of the potential well .
therefore one has that @xmath456 ( see columns 5 in table [ table_fluid_sub ] ) and @xmath457 ( see column 7 in table [ table_fluid_prop ] ) .
we evaluate the cross interaction potential parameters @xmath458 and @xmath459 based on kong s combining rules , because of their appreciated accuracy @xcite [ kong_rules ] @xmath460^{1/2},\ ] ] @xmath461^{1/13}\right\}^{13}.\ ] ] after all that we are ready to compute the key parameter @xmath462 that influences the behavior of the force for moderate values of @xmath0 .
the obtained results are reported in column 8 of table [ table_fluid_sub ] ) .
one observes that , depending on the choice of materials , one can indeed have both positive and _ negative _ parameters @xmath462 .
the existence of combination of materials for which @xmath462 is negative represents the main prerequisite for the experimental relevance of the effects predicted in the current article .
although the mean - field theory gives poor quantitative estimation of the behavior of the scaling function @xmath141 , it is tempting to , nevertheless , evaluate the corresponding force @xmath463 based on it .
to do so we consider the interaction between a cas planar substrate and a gold disk with radius @xmath464 and thickness @xmath465 , immersed in xenon .
we will assume that the separation between the confining surfaces is @xmath466 .
then for @xmath404 one shall observe double sign change of the force with the occurrence of two maxima , at which the force is repulsive and a minimum , at which it is attractive .
the magnitude of the force at the maxima is predicted to be @xmath467 and @xmath468 , observed at @xmath469 and @xmath470 respectively .
the minimum is found at @xmath471 and its value is @xmath472 .
finally we predict that the two sign changes occur at @xmath473 and @xmath474 respectively .
note that the predicted values of @xmath463 are significantly larger than the weight of the disk , which is approximately @xmath475 .
we stress that the predicted magnitude of the force is well in the range of the one reported in ref .
[ ] , where the authors estimated that an object with an effective interaction area @xmath476 at a distance @xmath477 to a wall experiences a force @xmath478 when immersed in a fluid with @xmath479 ( about the critical one of a water-2,6-lutidine mixture ) .
let us also mention that although the presented study is limited to the interaction between parallel planar surfaces through some fluid medium , it can be straightforwardly extended to objects of various geometries , e.g. , via using say the derjaguin or the gradient expansion approaches to calculate the interactions between curved objects .
then one can study also the behavior of colloidal particles immersed in a critical fluid or in a binary mixture close to its consulate point .
thus , another possible application of our findings is for governing the behavior of colloids in a critical solvent imagine coated colloid particles .
then , under proper choice of the core material of the colloid and of the coating the particles will have a stable equilibrium distance from the bounding plate that will depend on the temperature and the pressure of the fluid .
thus , we hope that our result can be used as a guide to the design of new experiments on colloid systems .
in the current article , in view of future experiments exploring and potential devices utilizing the critical casimir forces in classical one- or two - component fluids confined by parallel substrates at a distance @xmath0 from each other , we studied the behavior of the critical and the near critical casimir force and its interplay with the van der waals force , as well as the net force due to both of them . in the envisaged set - up the walls of the slabs
are considered coated by thin layers exerting strong preference to a given phase of the fluid modeled by strong adsorbing local surface potentials , while the slabs , on the other hand , influence the fluid by long - range dispersion potentials at least one of which supports the opposite phase of the fluid .
the fluid - fluid interactions are assumed to decay as @xmath480 , with @xmath87 being the distance between the particles of the medium , while the walls - fluid interactions decay as @xmath481 where @xmath89 is the fluid particle - single wall separation .
the strengths of the dispersion interactions in the system are depicted via the dimensionless coupling parameters @xmath220 and @xmath482 , accounting for the fluid - fluid and slabs - fluid interactions , respectively . while @xmath220 is always non - negative , for the slabs - fluid coupling parameters @xmath483 and @xmath413 we assumed that @xmath484 while @xmath485 .
the confined fluid medium is modeled within the framework of the lattice gas model defined in sec .
[ sec : model ] .
after developing a theory for arbitrary dimension @xmath9 and long - range decay exponent @xmath91 of the dispersion interactions see appendix [ sec : hamakerterm ] and [ sec : finsizebehav ] , we studied in sec .
[ sec : results ] the forces in @xmath84 systems , assuming that the dispersion interactions are of `` genuine '' van der waals type , i.e. , @xmath62 .
away from the critical region the total force @xmath98 between the plates of the bounded fluid is simply proportional to the hamaker constant see eq . and eq . , while near @xmath486 one has additional contribution to the force @xmath98 which is proportional to @xmath487 , where @xmath70 is a scaling function reflecting the role of the critical fluctuations of the order parameter see eq . .
using general scaling arguments , as well as explicit mean - field model calculations , we have obtained the following main results : \(1 ) in terms of some critical thickness @xmath2 of the thin films
( [ lcrit ] ) and ( [ cl1 ] ) , we have established a criterion for the importance of dispersion forces within the critical region of the system . for @xmath488
the contributions of the dispersion forces are important and can not be neglected neither within the critical region of the system , including the bulk critical point , nor outside of that region
[ fig : short_range_tending ] , [ fig : cfspsm ] and [ fig : cfsmsm ] where the validity of these statements is clearly visualized . in the opposite case ,
i.e. , when the separation @xmath0 between the walls is much larger than @xmath2 the dispersion forces provide only corrections to the leading behavior of the critical casimir and the total force within the critical region of the system . outside the critical region , however , i.e. , for @xmath489 , the influence of the dispersion ( van der waals ) interactions becomes essential for the behavior of the total force between the bounding surfaces .
this is of experimental importance because it is difficult to thermodynamically position the system right at the critical point .
\(2 ) in appendix [ sec : hamakerterm ] we have derived expressions for the behavior of the hamaker term in the total force between the plates see eq .
and eq . , that allow for a study of the temperature and the field dependencies of the hamaker constant including that ones within the critical region of the system .
the behavior of this part of the total force as a function of the scaling variables @xmath336 and @xmath337 for different values of @xmath490 , @xmath491 and @xmath220 is shown in fig .
[ fig : forceinterplay ] .
\(3 ) we demonstrated that for a suitable set of slabs - fluid and fluid - fluid coupling parameters the competition between the effects due to the coatings and the slabs can result in _ sign change _ of @xmath70 see figs .
[ fig : cfspsm ] and [ fig : cfsmsm ] , as well as of the casimir force
[ fig : forceinterplay ] , when one changes the temperature @xmath3 , the chemical potential of the fluid @xmath4 , or @xmath0 .
the last can be used in designing nano - devices and for governing behavior of objects at small , below micrometer , distances .
\(4 ) in addition to the critical behavior of the considered forces , in sec .
[ sec : phasebehaviour ] we discussed and gave detailed explanation for the near critical behavior of the phase diagram , see figs . [
fig : pddiffl ] , [ fig : pddiffsandlambda ] and [ fig : ncpd ] , of finite - size systems governed by dispersion interactions .
the influence of these interactions on the shift of the finite - size critical point has also been commented .
\(5 ) in the final sec .
[ sec : experimentalreal ] in scope of possible experimental realization of the theoretical set - up , we explored the usage of concrete substances . for each pair of substrate and fluid we calculated the coupling parameters @xmath441 and @xmath220 , and used this knowledge to make estimations for the critical casimir force @xmath463 at several temperatures where it exhibits extrema or changes its sign .
we have demonstrated that by proper choice of the fluid and substrate materials one can indeed achieve @xmath492 and @xmath493 , which is the main prerequisite for the experimental feasibility of the effects predicted in the current article .
let us stress that the problem of quantitative description of the mutual influence of the fluctuation of the electromagnetic field and the order parameter fluctuations of a medium when it is close to its critical point is extremely complicated and currently there is no general theory available to scope with it .
the lifshitz theory which is the basic one for studying the qed casimir effect has never been meant to nor can deal with the problem of a critical medium between two other media .
the main quantity which knowledge is required for practical applications of this theory is the dielectric permittivity @xmath494 .
it is normally tabulated at some temperature , usually the room temperature .
it shall be noted , however , that in a critical fluid @xmath494 is itself a singular function of the temperature @xcite .
we are not aware even of a theory that can reliably predict quantitatively how @xmath494 will depend on the temperature and @xmath495 near the critical point of the medium for a specific material characterized by some characteristic spectrum . on the other side
when studying critical phenomena one normally starts with some effective hamiltonian where only few basic features of the critical medium are reflected .
the current article provides a uniform treatment of the contributions due both to the van der waals forces and the critical casimir forces in the framework of an , unavoidably , relatively simple model in which , however , all the calculations have been done on equal footing .
the expression that we derived for the hamaker term is , nevertheless , in full agreement with the dzyaloshinskii - lifshitz - pitaevskii theory @xcite . on the other side
the mean - field theory is considered as a reliably theoretical horse for the qualitative description of the critical phenomena .
g. v. gratefully acknowledges the support of the bulgarian nsf under contract no .
dmu 03/37 .
d. d. thanks s. dietrich for helpful comments and the critical reading of the manuscript of this article .
the discussions with a. maciek are also gratefully acknowledged .
in this appendix we derive an expression for the hamaker term for a van der waals system of two different substrates separated by fluid medium .
we follow the same lines of action when performing the calculations as in ref .
@xcite with the important generalization that now , within our model ( see sec .
[ sec : model ] ) , the substrates on both sides of the fluid are allowed to be , in general , different from each other . where @xmath497 , and for the direct substrate - substrate potential we have assumed , in accordance with eqs .
( [ jldeftext ] ) and ( [ jlsdeftext ] ) , that it is given again by van der waals type interaction @xmath498 let us note that @xmath499 and @xmath500 , i.e. , the direct substrate - substrate interaction as well as the fluid - fluid interaction lead to attraction between the plates bounding the system while the fluid - substrate interactions are contributing to a _ repulsion _ between them .
furthermore , let us stress that @xmath67 and @xmath501 , are _ singular _ functions on the temperature near @xmath30 being dependent on the bulk order parameter @xmath227 , while @xmath64 is an analytic function of @xmath3 near @xmath502 .
using eqs .
( [ al ] ) ( [ alsi ] ) , for @xmath59 one immediately derives from eq . that @xmath503\right\}. \end{aligned}\ ] ] for the sake of simplicity in the next lines of this section we are going to omit giving explicitly the arguments of @xmath59 , @xmath504 and @xmath190 , as they do not change and are of no relevance for the following discussion . within our model @xmath67 and @xmath505 ,
@xmath203 being a part of the grand canonical potential @xmath506 have to be proportional to @xmath507 .
obviously , similar statement has to hold for @xmath64 .
thus , one has that this shall be also true for the pair potentials @xmath508 between any two substances @xmath509 and @xmath510 belonging to the system .
according to the general theory of the dispersion forces between molecules @xcite , neglecting the quantum effects , for the `` zero frequency contribution '' to @xmath511 one indeed has ( see eq .
( 6.24 ) on p. 121 in ref .
@xcite ) @xmath512 where @xmath513 and @xmath514 are the static electronic polarizabilities of each of the two substances @xmath509 and @xmath510 .
the last implies that the constants @xmath67 , @xmath515 , and @xmath64 are not independent . from eq .
one derives that @xmath516 using this equation one can rewrite eq . in the form given in eq . in the main text .
there , when calculating the critical component of the force , we often characterize the system by the values of the parameters @xmath220 and @xmath333 of the parameters @xmath104 at the bulk critical point of the fluid medium @xmath517 [ see eqs .
( [ deltav_s_l_def ] ) , ( [ l_param_def_new ] ) and ( [ s_def_l ] ) ] .
thus , it is convenient to express also @xmath59 in terms of these variables .
one obtains that @xmath518.\ \ \ \ \ \ \ \ \ \label{bethasigmaminusonehal}\end{aligned}\ ] ] since @xmath519 , with @xmath520 it is clear that @xmath428 , @xmath203 are temperature dependent , while @xmath333 are not . when @xmath110 eq
. simplifies to @xmath521 where we have taken into account that in mean - filed approximation @xmath173 . in eq .
the term proportional to @xmath522 corresponds to the direct interaction between the confining walls , the second ( proportional to @xmath523 ) corresponds to the interaction between the confining walls and the fluid medium , while the last one ( proportional to @xmath220 ) reflects the interaction between the constituents of the fluid medium .
the corresponding result for a @xmath62 system is @xmath524 which , taking into account eq .
, means that the hamaker constant is equal to @xmath525 .
\label{aham_d3sigma3_l}\ ] ] one must pay attention when using eqs .
( [ 3bethaha_ocf ] ) and ( [ 2bethaha_ocf ] ) , because the substrate - fluid coupling parameters @xmath333 depend on the dimensionality @xmath9 of the system and the decay exponent @xmath91 [ see eqs .
( [ deltav_s_l_def ] ) and ( [ s_def_l ] ) ] .
the behaviour of eq . as a function of the temperature and field scaling variables @xmath283 and @xmath331
is depicted in fig .
[ fig : forceinterplay ] . 168 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) .
, * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , eds . , _
_ , finite systems and multiparticle dynamics ( , ) . , _
_ ( , ) . , _ _
( , ) . , * * , ( ) . , , ,
* * , ( ) . , _ _ ( , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , _ _ ( , ) . , , , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , eds . , _ _ , no . in ( , ) . , , ,
* * , ( ) .
, , , * * , ( ) . , , , , , ,
* * , ( ) . ,
* * , ( ) . ,
( ) , . , * * , ( ) . , , ,
, eds . , _ _ , vol . of _ _ ( , ) , ed . , _ _
( , ) . , , , _ _ ( , ) . , * * , ( ) . ,
* * , ( ) .
, in _ _ , edited by ( , ) , p. .
, * * , ( ) . , *
, , , * * , ( ) . , _ _ ( , ) . , * * , ( ) , , , , , * * , ( ) . , * * , ( ) . , in _
_ , edited by ( , ) , vol . , p. . , _ _ ( , ) ,
vol . , chap . , pp . .
, , , , , * * , ( ) .
, * * , ( ) . , , , , * * , ( ) . , * * , ( ) .
, , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * ( ) . , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , _ _ ( , ) . , , , * * , ( ) .
, , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) .
, , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) .
, * * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , _ _
( , ) . , , ,
* * , ( ) . , , , * * , ( ) .
, , , , , * * , ( ) .
, , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , * * , ( . , * * , ( ) . , * * , ( ) .
, * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) .
, , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , * * , ( ) .
, , , * * , ( ) . , , , * * , ( ) .
, * * , ( ) . , * * , ( ) .
, , , * * , ( ) . , , , ,
* * , ( ) . ,
* * , ( ) . , _ _ ( , ) .
, , , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , ed . , _ _ ( , ) . , ed . , _ _ ( , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) .
, * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , _ _ ( , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , _ _ ( , ) . , * * , ( ) . , _ _ ( , ) . , _
_ ( , ) . ,
* * , ( ) .
, * * , ( ) . , , , , ( ) . , , , ,
, * * , ( ) . , ,
* * , ( ) .
, , * * , ( ) . , , * * , ( ) .
, , , , , * * , ( ) .
, , , * * , ( ) . , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) .
, , * * , ( ) . , , , , , , , * * , ( ) . , _ _ ( , ) . j. v. sengers , d. bedeaux , p. mazur , and s. c. greer , physica a * 104 * , 573 ( 1980 ) .
p. malik , s. j. rzoska , a. drozd - rzoska , and j. jadzyn , j. chem . phys .
* 118 * , 9357 ( 2003 ) . | we study , using general scaling arguments and mean - field type calculations , the behavior of the critical casimir force and its interplay with the van der waals force acting between two parallel slabs separated at a distance @xmath0 from each other , confining some fluctuating fluid medium , say a non - polar one - component fluid or a binary liquid mixture .
the surfaces of the slabs are coated by thin layers exerting strong preference to the liquid phase of the fluid , or one of the components of the mixture , modeled by strong adsorbing local surface potentials ensuring the so - called @xmath1 boundary conditions .
the slabs , on the other hand , influence the fluid by long - range competing dispersion potentials , which represent irrelevant interactions in renormalization group sense . under such conditions
one usually expects _ attractive _ casimir force governed by universal scaling function , pertinent to the extraordinary surface universality class of ising type systems , to which the dispersion potentials provide only corrections to scaling .
we demonstrate , however , that below a given threshold thickness of the system @xmath2 for a suitable set of slabs - fluid and fluid - fluid coupling parameters the competition between the effects due to the coatings and the slabs can result in _ sign change _ of the casimir force acting between the surfaces confining the fluid when one changes the temperature @xmath3 , the chemical potential of the fluid @xmath4 , or @xmath0 .
the last implies that by choosing specific materials for the slabs , coatings and the fluid for @xmath5 one can realize _ repulsive _ casimir force with _ non - universal _ behavior which , upon increasing @xmath0 , gradually turns into an _ attractive _ one described by an _
universal _ scaling function , depending only on the relevant scaling fields related to the temperature and the excess chemical potential , for @xmath6 .
we presented arguments and relevant data for specific substances in support of the experimental feasibility of the predicted behavior of the force .
it can be of interest , e.g. , for designing nano - devices and for governing behavior of objects , say colloidal particles , at small distances . we have formulated the corresponding criterion for determination of @xmath2 .
the universality is regained for @xmath6 .
we have also shown that for systems with @xmath5 the capillary condensation phase diagram suffers modifications which one does not observe in systems with purely short - ranged interactions . |
in a departure from classical voting theory , a growing literature from computational social choice has recently studied _ multi - winner _ voting rules : given diverse preferences of a collection of agents , instead of identifying a single best alternative , we are aiming for a ( fixed - size ) _ set _ of alternatives that jointly are able to represent the preferences of the agents best .
such procedures are useful in a wide variety of circumstances : obvious examples include the election of a parliament , or of a committee representing the interests of members of an organisation .
other applications can be found in group recommendation systems , or for making decisions about which products or services to offer : which courses should be offered at a university ? which movies should be presented on an airline entertainment system ? several attractive rules for such tasks have been designed by researchers in political science ( e.g. , @xcite @xcite , @xcite @xcite ) and more recently by computer scientists @xcite .
many of these rules are defined in terms of some objective function : a winning committee is a set of @xmath0 candidates that maximises this objective .
unsurprisingly , then , the winner determination problems of such rules are typically np - hard @xcite . to tackle the complexity of these problems , approximation algorithms @xcite and fixed - parameter tractability approaches @xcite have been developed for these problems , and integer programming formulations
have also been designed for them @xcite .
another approach to efficiently solving these winner determination problems seeks to exploit underlying structure in the preferences reported by the agents @xcite .
a particularly popular _ preference restriction _ in this space is the notion of _ single - peaked preferences _ , due to @xcite and @xcite . under this model ,
the alternative space has a one - dimensional structure : alternatives are ordered on a left - to - right _ axis _ ; and agents preferences are monotonically decreasing as we move further away from their _ peak _ ( most - preferred alternative ) . in particular , we can expect preferences to be structured this way when voting over the value of a numerical quantity ( such as a tax rate ) . while single - peaked preferences were first employed to escape impossibility results in social choice theory @xcite
, it also yields positive algorithmic results : notably , @xcite showed that chamberlin
courant s committee selection rule can be computed efficiently when preferences are single - peaked .
these results can be extended to other multi - winner voting rules @xcite , and to other preference restriction such as single - crossing preferences , or single - peakedness on trees @xcite .
the algorithms mentioned all directly exploit the underlying structure of the preferences .
for example , the algorithm due to betzler et al .
proceeds by dynamic programming along the single - peaked axis .
thus , to use this algorithm , we first need a procedure to uncover this axis .
fortunately , efficient recognition algorithms for the mentioned domains are known ; in particular , a suitable axis can be found in linear time @xcite .
thus , the approach to tractability via preference restrictions would look roughly like shown in : given input preferences , apply recognition algorithms for several preference restrictions , and if any of these tests succeed , use a special - purpose algorithm that performs well when preferences have the uncovered structure . if none of the tests succeed , fall back to some ( superpolynomial ) general - purpose solver , such as an integer programming solver .
but there is something awkward about linking together this cacophony of algorithms : given the amazing progress in solver technology in recent decades ( such as witnessed in sat competitions or in commercial ip solvers ) , it might well be faster in practice to skip these intermediate steps and go straight to cplex .
( this objection becomes stronger once implementation effort is factored in . )
further , modern ip solvers are often able to exploit underlying structure automatically : experimentally , solving times on single - peaked instances are much faster than on random instances .
while such an experimental result is nice , performance guarantees are better .
could it be that , on certain single - peaked instances , an ip solver performs exponentially worse than special - purpose algorithms ?
the answer is no .
this paper shows that integer programming solvers will _ provably _ terminate in polynomial time when solving certain voting problems on single - peaked inputs . in more detail , for several voting rules ( including chamberlin courant and proportional approval voting ) , we will design ip formulations which are solved optimally by their lp relaxation when the preference input happens to be single - peaked . since all standard ip solvers first solve the lp relaxation , they will terminate with the correct answer in their first iteration .
if the instance is not single - peaked , the ip solver might enter further iterations while solving
importantly , our formulations are correct whether or not the input is single - peaked .
moreover , this approach to achieving polynomial time efficiency does not require separately running a recognition algorithm !
the ip solver need not know an underlying axis , or even whether the input is single - peaked at all ; the lp relaxation will just ` magically ' have an integral solution .
our proofs rely on establishing that the constraint matrices become totally unimodular in the single - peaked case .
previous applications of this technique include certain tractable cases of the winner determination problems of combinatorial auctions ( see @xcite @xcite ) .
our method also allows us to show that a conjecture due to @xcite is false .
they consider structure in _ dichotomous preferences _ based on approval ballots , where agents only submit a binary yes / no decision for every candidate .
in particular , they consider an analogue of single - peakedness in this setting ( which they call _ candidate interval _ or _ ci _ ) , which requires there to be an ordering ( axis ) of the candidates such that , for every voter , their set of approved candidates forms an interval of the axis .
they then analyse an axiomatically particularly attractive multi - winner rule for the approval setting , known as proportional approval voting ( pav ) . while @xcite showed that pav is efficiently computable for a certain subclass of their ci concept , and obtained fpt results for ci preferences , they conjectured that pav remains np - hard for ci preferences in general .
this conjecture appears to be largely based on the difficulty of solving this problem using dynamic programming approaches that typically work in single - peaked settings .
our method , on the other hand , allows us to find a polynomial - time algorithm for this case , via ip solving .
we then combine the approaches for chamberlin courant and pav to give a similar polynomial time result for so - called owa - based multi - winner rules in the case that preferences are single - peaked .
we further give an ip formulation that computes young s voting rule efficiently for single-_crossing _ preferences , and close by briefly sketching applications of our method to some further rules and settings .
* total unimodularity * a matrix @xmath1 with @xmath2 is called _ totally unimodular _ if every square submatrix @xmath3 of @xmath4 has @xmath5 .
the following results are well - known .
proofs and much more about their theory can be found in the textbook by @xcite .
[ thm : tum - is - easy ] suppose @xmath6 is a totally unimodular matrix , @xmath7 is an integral vector of right - hand sides , and @xmath8 is an objective vector .
then the linear program @xmath9 has an integral optimum solution , which is a vertex of the polyhedron @xmath10 .
thus , the integer linear program @xmath11 can be solved using its linear programming relaxation ( p ) .
an optimum solution to ( ip ) can be found in polynomial time .
we will now state some elementary results about totally unimodular matrices .
[ prop : tum - manipulations ] if @xmath4 is totally unimodular , then so is 1 .
its transpose @xmath12 , 2 . the matrix @xmath13 $ ] obtained from @xmath4 by appending the negated columns of @xmath4 , 3 . the matrix @xmath14 $ ] where @xmath15 is the identity matrix , 4 . any matrix obtained from @xmath4 through permuting or deleting rows or columns . in particular , from ( 3 ) and ( 4 )
it follows that appending a unit column @xmath16 will not destroy total unimodularity .
further , using these transformations , we can see that remains true even if we add to ( p ) constraints giving lower and upper bounds to some variables , if we replace some of the inequality constraints by equality constraints , or change the direction of an inequality .
r0.23 a binary matrix @xmath17 has the _ strong consecutive ones property _ if the @xmath18-entries of each row form a contiguous block , as in the example on the right .
a binary matrix has the _ consecutive ones property _ if its columns can be permuted so that the resulting matrix has the strong consecutive ones property .
every binary matrix with the consecutive ones property is totally unimodular . by a celebrated result of @xcite
, it is possible to decide in polynomial time whether a given matrix is totally unimodular . + * * single - peaked preferences**let @xmath4 be a finite set of _ alternatives _ , or _ candidates _ , and let @xmath19 .
a _ weak order _ , or _
preference relation _ , is a binary relation @xmath20 over @xmath4 that is complete and transitive .
we write @xmath21 and @xmath22 for the strict and indifference parts of @xmath20 .
a _ linear order _ is a weak order that , in addition , is antisymmetric , so that @xmath23 only if @xmath24 .
every preference relation @xmath20 induces a partition of @xmath4 into indifference classes @xmath25 so that @xmath26 and @xmath27 for all @xmath28 .
we will say that an alternative @xmath29 has _ rank _
@xmath30 in the ordering @xmath20 and write @xmath31 ; thus the alternatives of rank 1 are the most - preferred alternatives under @xmath20 .
finally , we say that any set of the form @xmath32 is a _ top - inital segment _ of @xmath20 .
r0.4 let @xmath33 be ( the strict part of ) a linear order over @xmath4 ; we call @xmath33 an _ axis_. a linear order @xmath34 with most - preferred alternative @xmath35 ( the _ peak _ ) is _ single - peaked with respect to @xmath33 _ if for every pair of candidates @xmath36 with @xmath37 or @xmath38 it holds that @xmath39 . for example , if the alternatives in @xmath4 correspond to different proposed levels of a tax , and the numbers in @xmath4 are ordered by @xmath33 in increasing order , then it is sensible to expect voters preferences over @xmath4 to be single - peaked with respect to @xmath33 .
note that @xmath21 is single - peaked with respect to @xmath33 if and only if all top - initial segments of @xmath21 form an interval of @xmath33 .
accordingly , we will define a _ weak _ order @xmath20 to be _ single - peaked _ with respect to @xmath33 exactly if all top - initial segments of @xmath20 form an interval of @xmath33 .
this concept is often known as ` possibly single - peaked ' @xcite because it is equivalent to asking that all the ties in the weak order can be broken in such a way that the resulting linear order is single - peaked
. a _ profile _ @xmath40 over a set of alternatives @xmath4 is a list of weak orders over @xmath4 .
each of the orders represents the preferences of a _ voter _ ; we write @xmath41 $ ] for the set of voters .
the profile will be called _ single - peaked _ if there exists some axis @xmath33 over @xmath4 so that each order @xmath42 in @xmath43 is single - peaked with respect to @xmath33 .
a profile is single - peaked if and only if the following matrix @xmath44 has the consecutive ones property : take one column for each alternative , and one row for each top - initial segment @xmath45 of each voter s preference relation ; the row is just the incidence vector of @xmath45 .
this construction is due to @xcite .
+ * * dichotomous preferences**a weak order @xmath20 is _ dichotomous _ if it partitions @xmath4 into at most two indifference classes @xmath46 .
the alternatives in @xmath47 are said to be _ approved _ by the voter @xmath20 .
on dichotomous preferences , the notion of single - peakedness essentially coincides with the consecutive ones property @xcite : there needs to be an ordering @xmath33 of the alternative so that each approval set @xmath47 is an interval of @xmath33 .
thus , @xcite use the name _ candidate interval ( ci ) _ for single - peakedness in this context .
+ r0.265 * * single - crossing preferences**a profile @xmath48 of _ linear _ orders is called single - crossing if voters can be ordered so that for all @xmath36 , the set of voters who prefer @xmath49 to @xmath50 form an interval of this ordering . again , this can be phrased in terms of the consecutive ones property of a matrix @xmath51 built from the profile : take one column for each voter , and one row for each of the @xmath52 pairs @xmath36 ; there is a 1 in this row for each voter with @xmath53 .
in this section , we will consider proportional approval voting ( pav ) , a multiwinner voting rule defined for dichotomous ( approval ) preferences . a nave way to form a committee would be to select the @xmath0 alternatives with highest approval score , but this method tends to ignore minority candidates , and so is not representative @xcite .
pav attempts to fix this issue .
the rule appears to have been first proposed by @xcite . in the general case ,
a winning committee under pav is hard to compute @xcite .
let us define pav formally .
each voter @xmath54 submits a set @xmath55 of _ approved _ candidates ( or , equivalently , a dichotomous weak order with @xmath56 ) .
we aim to find a good committee @xmath57 of size @xmath58 .
the intuition behind proportional approval voting ( pav ) is that voters are happier with committees that contain more of their approved candidates , but that there are decreasing marginal returns to extra approved candidates in the committee . concretely
, each voter obtains a ` utility ' of 1 for the first approved candidate in @xmath59 , of @xmath60 for the second , of @xmath61 for the third , and so on .
the objective value of a committee @xmath57 is thus @xmath62 the choice of harmonic numbers might seem unusual , and one can more generally define a rule @xmath63 where @xmath64 is a non - increasing scoring vector ( so @xmath65 when @xmath66 ) .
this rule gives @xmath59 the objective value @xmath67 .
then pav is just @xmath68-pav .
however , the choice of harmonic numbers is the only vector @xmath69 that lets @xmath63 satisfy an axiom called ` extended justified representation ' @xcite , making this a natural choice after all .
let us now analyse our first ip formulation , for pav .
@xmath70 } \alpha_\ell \cdot x_{i,\ell } } & & \multicolumn{2}{r}{\text{(pav - ip ) } } \\ \text{subject to } & \displaystyle\sum_{c\in c } y_{c\hphantom{,\ell } } & = & k & & & \text{(2 ) } \\ & \displaystyle\sum_{\ell \in [ k ] } x_{i,\ell } & { \leqslant } & \displaystyle\sum_{\mathclap{i \text { approves } c } } \ :
y_c & & \text{for } i \in n & \text{(3 ) } \\ & x_{i,\ell } & \in & \{0,1\ } & & \text{for } i\in n,\ : \ell\in[k ] & \text{(4 ) } \\ & y_{c\hphantom{,\ell } } & \in & \{0,1\ } & & \text{for } c\in c \end{array}\ ] ] program correctly computes an optimal committee according to @xmath63 . in any feasible solution of , the variables @xmath71
encode a committee of size @xmath0 .
fix such a committee fix such a committee @xmath72 .
we show that in optimum , the objective value of the solution is the @xmath63-value of this committee . since @xmath73 , in optimum ,
as many @xmath74 will be set to 1 as constraint ( 3 ) allows .
thus , for each @xmath54 , exactly @xmath75 many variables @xmath76 will be set to 1 .
since @xmath77 is non - increasing , wlog , in optimum , these will be variables @xmath78 .
then the objective value equals the @xmath63-value of @xmath59 . an interesting feature of ( pav - ip ) is that the integrality constraints ( 4 ) on the variables @xmath74 can be relaxed to just @xmath79 ; this does not change the objective value of the optimum solution .
this is because ( a ) in optimum , the quantity @xmath80 is integral and ( b ) it never pays to then have one of the @xmath76 to be fractional , because this fractional amount can be shifted to @xmath81 with ( weakly ) higher value .
this observation might tell us that solving ( pav - ip ) is relatively easy as it is `` close '' to being an lp , and this also seems to be true in practice . on the other hand , it is not necessarily beneficial to relax the integrality constraints ( 4 ) when passing ( pav - ip ) to an ip solver : the presence of integrality constraints might nudge the solver to keep numerical integrality gaps smaller . of course , the point of this paper is to give another reason why ( pav - ip ) is `` close '' to being an lp .
[ prop : pav - tum ] the constraint matrix of is totally unimodular when the input preferences are single - peaked . we will use the manipulations allowed by proposition [ prop : tum - manipulations ] liberally .
in particular , it allows us to ignore the constraints @xmath82 , and to ignore the difference between equality and inequality constraints .
thus , after permuting columns corresponding to variables @xmath76 so that they are sorted by @xmath54 , the constraint matrix of ( pav - ip ) is @xmath83[l]{$\smash{\overbrace{\phantom { \begin{matrix}-i_n & \dots & -i_n\end{matrix}}}^{\text{$k$ times}}}$}-i_n & \dots & -i_n } & m_{\text{sp}}^p \\
\mathbf 0_n & \dots & \mathbf 0_n & \mathbf 1_m \end{bmatrix}.\ ] ] if preferences @xmath43 are single - peaked , then @xmath44 has the consecutive ones property , and this is also true after appending a row with all-1s .
thus , @xmath84 $ ] is totally unimodular . applying proposition [ prop : tum - manipulations ] repeatedly to append negations of unit columns , we obtain @xmath85 , which is thus totally unimodular .
using , we obtain our desired result .
@xmath63 can be computed in polynomial time for single - peaked approval preferences .
now we leave the domain of dichotomous preferences , and consider the full generality of profiles of weak orders . the definition of chamberlin
courant s rule is based on the notion of having a _ representative _ in the elected committee : each voter is represented by their favourite candidate in the committee , and voters are happier with more preferred representatives .
let @xmath86 be a ( non - increasing ) scoring vector ; the standard choice for @xmath77 are borda scores : @xmath87 .
let @xmath88 be a profile .
then the objective value of a committee @xmath57 according to chamberlin
courant s rule is @xmath89 chamberlin courant now returns any committee @xmath57 with @xmath58 that maximises this objective .
the rule thus defined can be seen as a ( non - metric ) facility location problem : each candidate @xmath90 is a potential facility location , we are allowed to open exactly @xmath0 facilities , and the distance between customers and facilities are determined through @xmath77 .
there is a standard integer programming formulation for this problem using binary variables @xmath71 , denoting whether @xmath91 will be opened or not , and variables @xmath92 , denoting whether facility @xmath91 will service voter @xmath54 .
we will need an alternative formulation based on maximising a number of _
points_. for expositional simplicity , let s take @xmath77 to be borda scores ; other scoring rules can be obtained by weighting the points .
here is another way of thinking about the objective value as defined above : each voter @xmath54 can earn a point for each _ rank _ in @xmath54 s preference order : for every rank @xmath93 $ ] , @xmath54 earns the point @xmath94 if there is a committee member @xmath95 with @xmath96
. then the number of points obtained in total equals the objective values : if @xmath54 s favourite committee member is in rank @xmath97 , then @xmath54 will earn precisely @xmath98 points , namely the points @xmath99 .
this view suggests the following integer programming formulation , where we put @xmath100 and @xmath101 .
@xmath102 } w'_r \cdot x_{i , r } } & & \multicolumn{2}{r}{\text{(cc - ip ) } } \\ \text{subject to } & \displaystyle\sum_{c\in c } y_{c } & = & k & & & \text{(2 ) } \\ & x_{i , r } & { \leqslant } & \displaystyle\sum_{\mathclap{c \ : : \ : { \textrm{rank}}(c ) { \geqslant}r } } \ : y_c & & \text{for } i \in
n,\ : r\in[m ] & \text{(3)}\\ & x_{i , r } & \in & \{0,1\ } & & \text{for } i\in n,\ : r\in[m ] \\ & y_{c\hphantom{,\ell } } & \in & \{0,1\ } & & \text{for } c\in c \end{array}\ ] ] program correctly computes an optimal committee according to @xmath77-chamberlin courant . in any feasible solution of , the variables @xmath71
encode a committee of size @xmath0 .
fix such a committee @xmath72 .
we show that in optimum , the objective value of the solution is the objective value of this committee according to @xmath77-cc . since @xmath103 , in optimum
, every @xmath94 will be set to @xmath18 if constraint ( 3 ) allows this .
this is the case iff there is a committee member @xmath95 with @xmath96 , i.e. , iff the ` point ' @xmath94 is earned as described above .
then it is clear that the objective of ( cc - ip ) corresponds to chamberlin
courant s objective .
the constraint matrix of is totally unimodular when the input preferences are single - peaked .
after similar simplification as in proposition [ prop : pav - tum ] we see that the constraint matrix of ( cc - ip ) is @xmath104 again , if preferences @xmath43 are single - peaked , then @xmath44 has the consecutive ones property , and this is also true after appending a row with all-1s .
thus , @xmath84 $ ] is totally unimodular . applying proposition [ prop : tum - manipulations ] repeatedly to append unit columns , we obtain @xmath105 , which is thus totally unimodular .
chamberlin courant with score vector @xmath77 can be solved in polynomial time for single - peaked preferences .
in the philosophy behind chamberlin courant , each voter is represented by exactly one committee member , and obtains all ` utility ' through this representation . in many application scenarios ,
we may instead seek _ multirepresentation _ @xcite : for example , you might watch several of the movies offered by an inflight - entertainment system . in such scenarios ,
chamberlin courant might design a suboptimal committee ; skowron et al .
introduce _ owa - based _ multiwinner rules as a more flexible alternative ( see also @xcite @xcite ) . given a vector @xmath106
, a weight vector @xmath107 defines an _ ordered weighted average _ ( owa ) operator as follows : first , sort the entries of @xmath108 into non - increasing order , so that @xmath109 ; second , apply the weights : the ordered weighted average of @xmath108 with weights @xmath110 is given by @xmath111 .
for example , @xmath112 gives the maximum , and @xmath113 gives the sum of the numbers in @xmath108 .
now , a scoring vector @xmath86 and an owa @xmath69 define an owa - based multi - winner rule as follows : given a profile @xmath43 , the rule outputs a committee @xmath114 that maximises the objective value @xmath115 thus , choosing @xmath116 gives us @xmath77-chamberlin courant as a special case .
choosing @xmath117 gives us an analogue of chamberlin courant where voters are represented by their favourite _ two _ members of the committee .
the owa - based rules with @xmath118 and @xmath119 gives us pav , when given dichotomous preferences as input .
thus , owa - based rules generalise both chamberlin courant and pav , and it turns out that we can apply our method to these rules by merging the ideas of ( pav - ip ) and ( cc - ip ) .
however , our formulation is only valid for _ non - increasing _ owa vectors with @xmath65 whenever @xmath66 .
thus , for example , we exclude the rule where voters are represented by their _ least_-favourite committee member . that are zero except for constantly many values at the end .
] for the ip , we again put @xmath100 and @xmath101 .
@xmath120 } } \sum_{\ell \in [ k ] } \sum_{r\in [ m ] } \alpha_\ell \cdot w_r ' \cdot x_{i,\ell , r } \hfill \text{(owa - ip ) } } \\ \text{subject to } & \displaystyle\sum_{c\in c } y_{c\hphantom{,\ell , r } } & = & k & & & \text{(2 ) } \\ & \displaystyle\sum_{\ell \in [ k ] } x_{i,\ell , r } & { \leqslant } & \displaystyle\sum_{\mathclap{c \ : : \ : { \textrm{rank}}(c ) { \geqslant}r } } \ : y_c & & \text{for } i \in
n,\ : r\in[m ] & \text{(3)}\\ & x_{i,\ell , r } & \in & \{0,1\ } & & \text{for } i\in n,\ : \ell , \ : r & \text{(4 ) } \\ & y_{c\hphantom{,\ell , r } } & \in & \{0,1\ } & & \text{for } c\in c \end{array}\ ] ] if @xmath69 and @xmath77 are non - increasing , correctly computes an optimal committee according to the owa - based rule based on @xmath69 and @xmath77 .
similarly to previous arguments , in optimum , we will have @xmath121 if and only if the committee @xmath72 contains at least @xmath122 candidates that voter @xmath54 ranks in rank @xmath97 or above .
thus , the objective value of ( owa - ip ) agrees with the defined objective of the owa - based rule .
the following property is proved very similarly to before .
the constraint matrix of is totally unimodular when input preferences are single - peaked .
owa - based rules with non - increasing owa operator can be solved in polynomial time for single - peaked preferences .
in contrast to the rules we have considered so far , young s voting rule does not select a committee ; it is a _ single_-winner voting rule that is based on extending condorcet s rule . given a profile @xmath43 , an alternative @xmath90 is a _ condorcet winner _ if it beats every other alternative in a pairwise majority contest , i.e. , for every @xmath123 , the number of voters in @xmath43 with @xmath124 strictly exceeds the number of voters with @xmath125 ( thus defined , condorcet winners are unique ) . as is well - known , not every profile admits a condorcet winner . how should we choose the outcome in such profiles ?
young s rule chooses the condorcet winners of the largest subprofiles that do admit a condorcet winner , where a subprofile can be obtained by _ deleting voters_. in other words , young s rule yields all alternatives that can be made condorcet winners by deleting a minimum number of voters .
young s rule also assigns to each alternative @xmath91 a _ score _ , namely the number of voters in a maximum subprofile of @xmath43 in which @xmath91 is the condorcet winner ; young winners are the alternatives with maximum score .
deciding whether a given alternative is a young winner is @xmath126-complete @xcite , but is solvable in polynomial time for single - peaked @xcite and single - crossing preferences , essentially because such profiles always admit a condorcet winner ( in odd electorates ) . since , in general , it is hard to find a young winner , it is also hard to find the young _ score _ of a given alternative @xmath49 .
this latter problem admits the following simple ip formulation . here
, the variables @xmath127 indicate whether voter @xmath54 is to be deleted from the profile , and @xmath128 is the _ majority margin _ of @xmath50 over @xmath49 .
@xmath129 while previous ips were easy for single - peaked profiles , this one behaves nicely for single-_crossing _ preferences : the constraint matrix of is totally unimodular when input preferences are single - crossing . after disregarding the constraints @xmath130 ,
the constraint matrix is a submatrix of @xmath51 , consisting of just the rows corresponding to pairs @xmath131 for @xmath132 .
if input preferences are single - crossing , then @xmath51 has the consecutive ones property , and hence so do all its submatrices .
it follows that young scores can be computed in polynomial time when preferences are single - crossing .
this in itself is not an impressive result , since there is a very easy direct algorithm for solving this special case , on top the median yields the young score of @xmath49 .
see also @xcite .
] but the advantage of the method via integer programming is that we need not check whether @xmath43 is single - crossing or not to obtain the efficiency gains .
* more than single - peakedness . *
our polynomial - time results apply to a slightly larger class than just single - peaked profiles : they also apply when @xmath133 ( with an all-1s row appended ) is totally unimodular but does not necessarily have the consecutive ones property .
it can be shown that this is the case whenever @xmath43 contains only two distinct voters , or , more generally , when the set of all top - initial segments of @xmath43 can also be induced by a two - voter profile .
together with single - peaked profiles , we conjecture that these classes of profiles are precisely the profiles for which the relevant constraint matrices are totally unimodular .
* egalitarian versions .
* we can obtain egalitarian versions of the multi - winner rules that we have discussed by replacing the sum over @xmath134 by a minimum in their objective values @xcite . for pav and chamberlin courant
, our ip formulations can easily be adapted to answer the question `` is there a committee with egalitarian objective value @xmath135 ? '' while preserving total unimodularity in the case of single - peaked preferences .
an optimum committee can then be found by a binary search on @xmath136 .
however , it is unclear how this can be achieved for owa - based rules .
it is also unclear how to handle other utility aggregation operators such as leximin ( see @xcite @xcite ) .
* pav and voter intervals . * @xcite define an analogue of single - crossingness for dichotomous preferences called _ voter interval _
( vi ) , which requires the _ transpose _ of @xmath133 to have the consecutive ones property . as for ci , they conjectured that pav remains hard on vi preferences .
we could not solve this problem using our method : the constraint @xmath137 of ( pav - ip ) destroys total unimodularity .
* dodgson s rule . *
an alternative is a
_ dodgson winner _ if it can be made a condorcet winner using a minimum number of swaps of adjacent alternatives .
this number of swaps is the _ dodgson score _ of an alternative . @xcite
give an ip formulation for this problem , which is also used in the treatment of @xcite .
sadly , while ` most ' of the constraint matrix is again identical to @xmath133 , some extra constraints ( saying that the swaps in each vote should only count once ) destroy total unimodularity , so our method can not be employed for this formulation . note
that while @xcite give an efficient algorithm for finding a dodgson _ winner _ in the case of single - peaked preferences , the problem of efficiently calculating _ scores _ appears to be open and non - trivial .
maybe there is an alternative ip formulation that can be made to work using our approach
. * kemeny s rule .
* @xcite present several ip formulations for kemeny s rule .
the poly - size formulation they give involves constraints enforcing transitivity of the kemeny ranking ; these constraints are not totally unimodular . in any case
, most strategies for calculating kemeny s rule first calculate all pairwise majority margins ; we might as well check for transitivity at this stage trying to use fancy total unimodularity is unnecessary .
aziz , h. ; brill , m. ; conitzer , v. ; elkind , e. ; freeman , r. ; and walsh , t. 2015a .
justified representation in approval - based committee voting . in _ proceedings of the twenty - ninth aaai conference on artificial intelligence _
, 784790 .
aaai press .
bredereck , r. ; faliszewski , p. ; niedermeier , r. ; skowron , p. ; and talmon , n. 2015 .
elections with few candidates : prices , weights , and covering problems . in _ international conference on algorithmic decision theory _ , 414431 .
springer .
caragiannis , i. ; covey , j. a. ; feldman , m. ; homan , c. m. ; kaklamanis , c. ; karanikolas , n. ; procaccia , a. d. ; and rosenschein , j. s. 2009 . on the approximability of dodgson and young elections . in
_ proceedings of the twentieth annual acm - siam symposium on discrete algorithms _ , 10581067 .
society for industrial and applied mathematics .
elkind , e. ; lackner , m. ; and peters , d. 2016 .
preference restrictions in computational social choice : recent progress . in _ proceedings of the twenty - fifth international joint conference on artificial intelligence ,
ijcai 2016_. faliszewski , p. ; hemaspaandra , e. ; hemaspaandra , l. a. ; and rothe , j. 2011 .
the shield that never was : societies with single - peaked preferences are more open to manipulation and control .
209(2):89107 .
faliszewski , p. ; skowron , p. ; slinko , a. ; and talmon , n. 2016a .
multiwinner analogues of the plurality rule : axiomatic and algorithmic perspectives . in _ proceedings of the thirtieth aaai conference on artificial intelligence_. faliszewski , p. ; skowron , p. ; slinko , a. ; and talmon , n. 2016b .
committee scoring rules : axiomatic classification and hierarchy . in _ proceedings of the 25th international joint conference on artificial intelligence ( ijcai-2016)_. skowron , p. ; faliszewski , p. ; and lang , j. 2015 . finding a collective set of items : from proportional multirepresentation to group recommendation . in
_ proceedings of the twenty - ninth aaai conference on artificial intelligence _ , 21312137 .
aaai press . | _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ many np - hard winner determination problems admit polynomial - time algorithms when restricting inputs to be single - peaked .
commonly , such algorithms employ dynamic programming along the underlying axis .
we introduce a new technique : carefully chosen integer linear programming ( ip ) formulations for certain voting problems admit an lp relaxation which is totally unimodular if preferences are single - peaked , and which thus admits an integral optimal solution .
this technique gives fast algorithms for finding optimal committees under the pav and chamberlin courant voting rules under single - peaked preferences , as well as for certain owa - based rules . under single - crossing preferences ,
young scores can also be calculated .
an advantage of this technique is that no special - purpose algorithm needs to be used to exploit structure in the input preferences : any standard ip solver will terminate in the first iteration if the input is single - peaked , and will continue to work otherwise .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 1 @topnum0 @botnum0 |
* the jones polynomial . *
the jones polynomial @xmath5 of an oriented link @xmath6 in @xmath7 is determined by the skein relation @xmath8 for any three links @xmath9 that differ as shown in figure [ skein ] , and its value on the unknot , which we choose to be @xmath10 [ htb ] the jones polynomial does not depend on the framing of link components , but depends slightly on their orientations . reversing the orientation of a component @xmath11 multiplies the polynomial by @xmath12 where @xmath13 is the linking number of @xmath14 with its complement in @xmath15 * the colored jones polynomial .
* we briefly recall the basics about the colored jones polynomial of links ( for more details consult @xcite , @xcite , ( * ? ? ?
* sections 3,4 ) , and references therein ) . given an oriented framed link @xmath6 whose
components are colored ( marked ) by non - negative integers ( or , equivalently , by irreducible representations of @xmath16 with integer @xmath0 corresponding to the @xmath2-dimensional representation @xmath17 ) , the colored jones polynomial @xmath18 takes values in @xmath19.$ ] the label @xmath20 stands for the coloring , that is , for a function from the set of components of @xmath6 to non - negative integers .
if all components of @xmath6 are colored by @xmath21 the invariant is the original jones polynomial of @xmath15 if a component of @xmath6 is colored by @xmath22 deleting this component preserves the value of the colored jones polynomial . for a framed knot @xmath23
we denote by @xmath24 the jones polynomial of @xmath23 colored by @xmath25 thus , @xmath26 is the original jones polynomial of @xmath23 ( and does not depend on the framing ) , while @xmath27 for any knot @xmath28 more generally , one could label link components by arbitrary finite - dimensional representations of @xmath29 this does not give any extra information since the invariant is additive relative to the direct sum of representations , @xmath30 but allows us to express the colored jones polynomial via the jones polynomials of cables .
first , the jones polynomial @xmath31 of a knot @xmath23 labelled by @xmath32 equals the jones polynomial of the 2-cable @xmath33 of @xmath23 with components labelled by @xmath34 and @xmath35 the jones polynomial @xmath36 equals the original jones polynomial of the @xmath0-cable @xmath37 of @xmath28 for generic @xmath38 the quantum group @xmath39 has one irreducible representation @xmath17 in each dimension ( we consider representations where @xmath40 has only powers of @xmath38 as eigenvalues ) .
the grothendieck group of this category of representations is free abelian with a basis given by images @xmath41 , n\ge 0 $ ] of all irreducible representations .
we put square brackets around @xmath17 to distinguish a representation from its image in the grothendieck group .
another basis in the grothendieck group is given by all tensor powers of @xmath42 using the formula @xmath43 and induction on @xmath44 one checks that @xmath45 = \sum_{k=0}^{\lfloor\frac{n}{2}\rfloor } ( -1)^k\sbinom{n - k}{k } [ v_1^{\otimes ( n-2k)}].\ ] ] thus , the colored jones polynomial @xmath24 of a knot @xmath23 can be expressed via jones polynomials of its cables : @xmath46 this formula generalizes from knots to links , by summing over all connected components of a link : @xmath47 where @xmath48 is a function from the set of link components to non - negative integers .
we define the colored jones polynomial by formula ( [ def - color ] ) .
since the jones polynomial ( which appears on the right ) depends slightly on the orientation of each cable component , we need to specify our choices of orientations .
each component of @xmath6 is oriented .
when forming the @xmath49-cable of a component , if @xmath49 is even we orient half of the strands one way and the rest the opposite way .
if @xmath49 is odd , we orient @xmath50 strands by way of the original orientation of the component , and the remaining strands the opposite way . with this definition ,
@xmath18 does not depend on the orientations of components colored by even integers , while reversing the orientation of an odd - colored component @xmath14 multiplies the polynomial by @xmath51 , where @xmath52 is the sublink of @xmath6 made of all odd - colored components . if @xmath23 is the 0-framed unknot , @xmath53 { \stackrel{\mbox{\scriptsize{def}}}{=}}\frac{q^{n+1 } - q^{-n-1}}{q - q^{-1}},\ ] ] the quantum dimension of @xmath54 changing the framing of a component multiplies the colored jones polynomial by a power of @xmath55 as explained in the formula below and figure [ curl ] .
[ htb ] @xmath56 where @xmath0 is the color of the component .
* reduced colored jones polynomial . *
another common normalization for the colored jones polynomial of knots is @xmath57}.\ ] ] we will call @xmath58 the _ reduced _ colored jones polynomial .
the reduced polynomial of the @xmath59-framed unknot is @xmath4 we extend this normalization to colored links with a distinguished component , by @xmath60},\ ] ] where @xmath0 is the color of the distinguished component .
the reduced colored jones polynomial has integral coefficients .
* cohomology of cables . * take a framed link @xmath6 , its cable @xmath61 and consider the complex @xmath62 and its cohomology groups @xmath63 ( we refer the reader to @xcite for definitions of @xmath64 and @xmath65 ) . these groups are invariants of @xmath66 and depend non - trivially on its framing .
the euler characteristic of @xmath63 is the jones polynomial of the cable @xmath67 we will use the grading conventions of @xcite .
* categorification of @xmath68 * we would like to have a cohomology theory of links whose euler characteristic is the colored jones polynomial . let s start with the first non - trivial example beyond the original jones polynomial : our link is a framed knot @xmath23 colored by the three - dimensional representation @xmath69 note that @xmath70 is a direct summand of @xmath71 and the complementary summand is the trivial representation : @xmath72 this formula translates into a simple relation between the jones polynomials : @xmath73 to categorify @xmath74 we take the cohomology of the 2-cable of @xmath28 since @xmath75 we simply assume that its categorification is the abelian group @xmath76 placed in bidegree @xmath77 formula ( [ 2case ] ) says that to categorify @xmath78 we need to `` substract '' @xmath76 from @xmath79 this could be achieved by taking the cone of some map @xmath80 or of a map going in the opposite direction , from @xmath76 to @xmath79 this map should be natural . in the cohomology theory @xmath65
natural maps come from cobordisms between links .
note that @xmath76 is the cohomology of the empty link @xmath81 and there is a canonical cobordism in @xmath82 $ ] between @xmath33 and @xmath83 in the 2-cable @xmath33 two copies of @xmath23 run parallel next to each other , and there is a standard embedding of an annulus into @xmath7 with @xmath33 as its boundary ( the two components of @xmath33 will be oppositely oriented if we orient our annulus and induce the orientation onto its boundary ) .
push the interior of the annulus into the interior of @xmath82 $ ] slighly away from the boundary component @xmath84 that contains @xmath85 the resulting annulus @xmath86 in @xmath82 $ ] is an oriented cobordism between @xmath33 and the empty link .
its euler characteristic is @xmath59 and it induces a bidegree @xmath87 map between bigraded cohomology groups @xmath88 and @xmath89 this map is well - defined up to overall minus sign . on the level of complexes , we have a map @xmath90 well - defined ( up to the minus sign ) in the homotopy category of complexes .
we define the complex @xmath91 as the cone of @xmath92 shifted by @xmath93 $ ] ( so that @xmath76 is in degree @xmath94 ) , and cohomology groups @xmath95 as the cohomology of @xmath96 there is a short exact sequence of complexes @xmath97 \lra \cc_2(k ) \lra \cc(k^2 ) \lra 0\ ] ] giving rise to a long exact sequence of cohomology groups @xmath98 in this sequence at most one boundary map could be non - zero ( the one from @xmath76 to @xmath99 ) .
this map is non - zero if and only if @xmath100 is not surjective .
it s easy to see , and we do it below , that @xmath101 is a subgroup of index @xmath94 or @xmath102 in @xmath103 strictly speaking , in the above discussion we should fix a plane diagram @xmath104 of @xmath105 the associated diagram @xmath106 of the 2-cable @xmath107 and consider the complex @xmath108 from two diagrams @xmath109 and @xmath110 of @xmath23 related by a reidemeister move for framed knots , we obtain diagrams @xmath111 and @xmath112 of @xmath33 and a diagram of complexes and homomorphisms @xmath113 which commutes up to overall minus sign .
therefore , the cones of @xmath114 and @xmath115 are homotopy equivalent and isomorphism classes of groups @xmath116 are invariants of @xmath28 since @xmath86 is also a cobordism from the empty link to @xmath107 it induces a map @xmath117 and we could alternatively define the cohomology groups of @xmath23 colored by @xmath102 as the cohomology of the cone of @xmath118 ideally , we would like the two resulting cohomology theories to be naturally isomorphic .
right off , we could see that they are if @xmath102 is invertible in the base ring .
let @xmath119 be a commutative ring where @xmath102 is invertible ( for instance , a field of characteristic other than @xmath120 or @xmath121.$ ] ) tensoring @xmath122 and @xmath100 with @xmath119 we get maps @xmath123 and @xmath124 and the induced maps on cohomology groups ( we do nt bother inventing different notations for these maps ) .
the composition @xmath125 is the value of the invariant on the composition of the two cobordisms given by the annulus @xmath126 this composition is a cobordism in @xmath82 $ ] between empty links , and is isotopic to the torus standardly embedded into @xmath127 its invariant is @xmath128 therefore , @xmath129 since @xmath102 is invertible , we can decompose @xmath130 and derive isomorphisms @xmath131 proving that the two definitions lead to isomorphic cohomology theories ( @xmath132 in ( [ k - cong ] ) stands for cohomology of @xmath133 and @xmath134 in ( [ k - cong-2 ] ) for the cohomology of the total complex of @xmath135 ) .
the above argument implies that @xmath136 has index @xmath94 or @xmath137 _ framing : _ the colored jones polynomial @xmath78 depends in a simple way on the framing of @xmath28 change in the framing multiplies the polynomial by @xmath138 the cohomology theory @xmath95 has a more complicated dependence on the framing .
already when @xmath23 is the unknot with framing @xmath139 the rank of @xmath88 is @xmath140 and that of @xmath95 is @xmath141 in particular , cohomology of two knots that differ by a framing are not just overall shifts of each other . in general , if @xmath23 is a knot and @xmath142 is obtained from @xmath23 by twisting by a large slope , an overall shift superimposes @xmath95 and @xmath143 except for a long tail in @xmath143 trailing along two adjacent diagonals in the bigrading plane ( the tail is essentially the cohomology of the unknot with a large framing ) .
_ example : _ for the @xmath59-framed unknot @xmath105 @xmath144 for both definitions of @xmath145 _ note : _ the complex @xmath91 and its cohomology @xmath95 do not depend on the orientation of @xmath28 * categorification of @xmath146 * direct sum decomposition @xmath147 implies that , for a framed knot @xmath105 @xmath148 to categorify @xmath149 we form the @xmath150-cable @xmath151 of @xmath105 take its cochain complex @xmath152 and its cohomology @xmath153 choose a plane diagram of @xmath23 and the corresponding diagram of the cable .
enumerate the components of @xmath151 by @xmath154 and @xmath150 so that @xmath102 is in the middle , and orient component @xmath102 oppositely from @xmath94 and @xmath155 see figure [ cable3 ] .
[ htb ] consider two annuli in the neighbourhood of @xmath23 in @xmath156 one with components @xmath94 and @xmath102 as the boundary , and the other with components @xmath102 and @xmath150 as the boundary .
push the interiors of the annuli into @xmath82 $ ] away from the boundary .
two cobordisms from @xmath157 to @xmath23 result , inducing two maps from @xmath152 to @xmath158 and two maps on the cohomology , well - defined up to overall minus sign .
we can guess that the categorification of @xmath159 is the total complex of the bicomplex @xmath160 denote the total complex by @xmath161 and its cohomology by @xmath162 we get a bigraded cohomology theory of knots with @xmath159 as the euler characteristic .
independence of @xmath163 from the choice of planar diagram is straightforward . to categorify @xmath24 for arbitrary @xmath0 we should look for a bicomplex built out of complexes @xmath164 with binomial multiplicities as in ( [ def - knots ] ) , then form the total complex and take its cohomology .
this is done in section [ first - cat ] , while the representation - theoretic counterpart of this construction is worked out in the next section .
in this section we give a homological interpretation of the formula @xmath166 = \sum_{k=0}^{\lfloor\frac{n}{2}\rfloor } ( -1)^k\sbinom{n - k}{k } [ v_1^{\otimes ( n-2k)}]\ ] ] describing the image of @xmath17 in the grothendieck group via those of tensor powers of the defining representation @xmath42 we consider the @xmath167 case , so that @xmath17 s are representations of the lie algebra @xmath168 the case of generic @xmath38 is identical ( by substituting below `` @xmath38-antisymmetrization '' for `` antisymmetrization '' ) .
the right hand side of ( [ resol - formula ] ) has alternating coefficients , due to @xmath169 , and we could try to realize it as the euler characteristic of a complex .
we put the direct sum of @xmath170 copies of @xmath171 in the @xmath119-th cohomological degree and look for a natural differential @xmath172 the differential reduces the number of @xmath173 s in each tensor power by @xmath137 since @xmath174 there is a unique , up to scaling , surjective homomorphism @xmath175 and it is simply the antisymmetrization . we will use @xmath176 to construct the differential .
the binomial coefficient @xmath170 equals the number of ways to select @xmath119 pairs of neighbours from @xmath0 dots placed on a line , such that each dot appears in at most one pair ( the @xmath177 example is depicted in figure [ pairings ] ) .
we will call these @xmath119-pairings .
[ htb ] notice that the exponent in the tensor power of @xmath173 is the number of dots without a partner ( `` single '' dots ) .
the differential goes in the direction of increasing the number of pairs by @xmath4 a new pair will consists of two adjacent dots , and the differential should contract the two corresponding powers of @xmath173 into the trivial representation . to formalize , let @xmath178 be the set of @xmath119-pairings of @xmath0 dots .
for @xmath179 let @xmath180 be the set of single dots in @xmath181 and @xmath182 be the tensor product of @xmath173 s , one for each single dot .
if @xmath183 contains @xmath184 ( each pair in @xmath184 is also a pair in @xmath185 ) , there is a map @xmath186 given by contracting the two copies of @xmath173 representing the only pair in @xmath187 consider a graph with with vertices elements of @xmath188 over all @xmath189 and arrows
inclusions @xmath190 as above . for @xmath191
this graph is depicted in figure [ graph ] .
assigning @xmath192 to the vertex @xmath184 and the map @xmath193 to the arrow from @xmath184 to @xmath185 we obtain a commutative diagram of @xmath165 representations .
[ htb ] we make each square in the diagram anticommute by switching from @xmath193 to @xmath194 where @xmath195 is the number of pairs in @xmath184 to the left of the only pair in @xmath187 for each @xmath119 take the direct sum of @xmath196 for all @xmath197 and the sum of maps as above .
the result is a complex , denoted @xmath198 _ example : _ the complex @xmath199 has the form @xmath200 with the differential the sum of arrows in figure [ graph ] , the top left arrow appearing with the minus sign .
the complex @xmath201 is acyclic in non - zero degrees and its degree @xmath59 cohomology is the irreducible @xmath165 representation @xmath54 _ proof : _
@xmath202 is the subrepresentation of @xmath203 which is the kernel of @xmath204 this map is the sum of antisymmetrizations of two @xmath173 s over all pairs of neighbours .
therefore , @xmath202 is isomorphic to the @xmath0-th symmetric power of @xmath205 and to @xmath54 to prove exactness everywhere else , proceed by induction on @xmath0 .
the short exact sequence @xmath206 \lra c_n \lra c_{n-1 } \otimes v_1 \lra 0\ ] ] comes from separating all pairings of @xmath0 dots into two types : the one where the leftmost dot belongs to a pair , and the one where it does not .
the sum of all first type selections is a subcomplex of @xmath201 isomorphic to @xmath207 shifted one degree to the right , the sum of all second type selections is a quotient complex of @xmath201 isomorphic to @xmath208 induction hypothesis and the long exact sequence of ( [ short - exact ] ) imply @xmath209 for @xmath210 and exactness of @xmath211 ) \lra h^1(c_n ) \lra 0\ ] ] substituting @xmath212 for @xmath213 we get a short exact sequence @xmath214 telling us that @xmath215 @xmath216 _ remark : _ @xmath201 is a resolution of a simple module by a complex of semisimple modules .
this is different from the homological algebra framework , where we usually resolve a module by a complex of projective , or injective , or flat modules .
the category of finite - dimensional @xmath165 representations is already semisimple , all additive functors from this category are exact , and there is no need for resolutions from the homological algebra viewpoint .
however , @xmath201 seems to be interesting on it own .
* in characteristic @xmath137 * to avoid the sign ambiguity , in this subsection we work in characteristic @xmath137 let @xmath217 be the 2-element field . for a diagram @xmath104 of a link
, we denote the complex @xmath218 by @xmath219 and its cohomology by @xmath220 start with a framed oriented knot @xmath23 and its plane diagram @xmath221 it gives rise to diagrams @xmath222 of cables @xmath223 the cables are oriented as in figure [ orient ] , and , in a cross - section of @xmath224 orientations alternate .
choose a cross - section of @xmath222 and enumerate the strands from left to right from @xmath94 to @xmath0 so that component @xmath94 is oriented in the same way as @xmath225 component @xmath102 is oppositely oriented , etc .
[ htb ] for a pairing @xmath179 denote by @xmath226 the cable diagram containing only components corresponding to single dots ( not in any pair ) in @xmath181 and by @xmath227 the corresponding link ( it has @xmath228 components ) . given an arrow @xmath229
there is a canonical cobordism @xmath230 from @xmath227 to @xmath231 given by `` contracting '' the pair @xmath232 of neighbouring components of @xmath226 via an annulus .
this cobordism induces a well - defined ( up to homotopy ) map @xmath233 of complexes , and the induced map on cohomology ( also denoted @xmath234 ) let @xmath235 be the complex which in the @xmath119-th degree is the direct sum of @xmath236 over all @xmath237 the differential is the sum of @xmath193 over all possible arrows @xmath238 denote by @xmath239 the cohomology of @xmath240 these groups are bigraded , @xmath241 if diagrams @xmath104 and @xmath242 are related by a reidemeister move ii or iii , the complexes @xmath235 and @xmath243 are isomorphic .
this implies isomorphism classes of bigraded groups @xmath239 do not depend on the diagram @xmath104 of a framed knot @xmath105 and are invariants of @xmath28 their euler characteristic is the colored jones polynomial @xmath244 _ remarks : _ these groups do not depend on the orientation of @xmath28 _ from knots to links . _ given a colored link @xmath245 we choose its diagram @xmath225 then do the above procedure for each component of @xmath15 the result is a multi - dimensional commutative diagram of groups @xmath246 with multiplicities products of binomials , for various pairings @xmath247 and associated cables @xmath248 of @xmath221 note that @xmath249 and the pairing @xmath250 where @xmath251 is a pairing of @xmath252 dots ( equivalently , of @xmath252 cables of the @xmath253-th component ) .
we collapse this commutative diagram into a complex , denoted @xmath254 ( in characteristic @xmath102 a commutative square is anticommutative ) .
its cohomology groups do not depend on @xmath225 and are invariants of @xmath15 their euler characteristic is @xmath255 * cobordisms . * a cobordism in @xmath256 between colored framed links can be represented by a sequence of its cross - sections , each a generic plane diagram of a colored framed link , with each two consequent cross - sections related by either a reidemeister ii or iii move , or by a morse move , see figure [ morse ] .
components involved in a saddle point move should be colored by the same number , see figure [ morse ] .
[ htb ] a reidemeister ii or iii move from @xmath104 to @xmath109 induces an obvious isomorphism of groups @xmath246 and @xmath257 and of complexes @xmath254 and @xmath258 the unit and counit maps of @xmath259 induce natural maps between the complexes for the empty link and for the crossingless unknot diagram colored by @xmath25 we assign these maps to the `` birth '' and `` death '' morse moves .
suppose that diagrams @xmath104 and @xmath109 are related by a saddle point move .
consider the case when the move merges two components of @xmath104 ( both labelled by @xmath0 and denoted @xmath142 and @xmath260 ) into one component @xmath23 of @xmath261 see figure [ slab ] .
[ htb ] we would like to come up with a natural map @xmath262 of complexes @xmath263 where @xmath264 is the coloring of @xmath109 induced by the coloring @xmath20 of @xmath221 as an abelian group , @xmath254 is the direct sum of @xmath246 over all pairings @xmath247 of @xmath265 given @xmath266 let @xmath267 and @xmath268 be the pairings of @xmath0 dots which are the restrictions of @xmath247 to the components @xmath142 and @xmath269 let @xmath270 be the number of pairs in @xmath267 and @xmath271 the number of pairs in @xmath272 [ htb ] if pairs in @xmath267 and @xmath268 have at least one common dot ( see figure [ common ] ) , we set @xmath273 otherwise , @xmath267 and @xmath268 are disjoint and their union @xmath274 is a @xmath275-pairing of @xmath0 dots ( see figure [ disjoint ] ) .
[ htb ] to such @xmath247 we assign a pairing @xmath276 of the @xmath264 cable of @xmath277 this pairing is the same as @xmath247 on components of @xmath109 that are unchanged during the saddle point move , and @xmath274 on the component @xmath23 colored by @xmath25 the map @xmath262 will take @xmath246 to @xmath278 a pair in @xmath268 connects two dots numbered @xmath49 and @xmath279 for some @xmath49 ( in the figure [ disjoint ] example @xmath280 ) . for each such pair
we apply the operator of multiplication by @xmath281 at strand @xmath49 of @xmath282 on the cohomology @xmath246 ( recall that @xmath281 is the generator of the ring @xmath283 and @xmath284 ) . likewise ,
for a pair in @xmath267 connecting two dots numbered @xmath285 and @xmath286 ( in the figure [ disjoint ] example @xmath287 ) , we apply the operator of multiplication by @xmath281 at strand @xmath285 of @xmath288 denote by @xmath289 the product of these operators .
@xmath289 is an endomorphism of @xmath290 a multiplication by @xmath291 copies of @xmath281 at certain @xmath291 strands of the cable @xmath292 for each pair in @xmath268 ( connecting dots / strands numbered @xmath49 and @xmath279 ) consider a thin annulus whose boundary is the union of two strands of @xmath282 labelled @xmath49 and @xmath293 similarly , for each pair in @xmath267 ( connecting two dots numbered @xmath285 and @xmath286 for some @xmath285 ) consider a thin annulus whose boundary is the union of two strands of @xmath294 labelled @xmath285 and @xmath286 .
figure [ merge2 ] depicts these annuli schematically for our example .
the resulting @xmath291 annuli give rise to a cobordism from the cabled link with diagram @xmath248 to a cabled link with @xmath295 fewer components , whose diagram @xmath242 can be produced by removing these @xmath295 components ( the `` mirror '' partners of pairs in @xmath267 and @xmath268 ) from @xmath292 this cobordism induces a map of cohomology groups from @xmath246 to @xmath296 denoted @xmath297 in the diagram @xmath242 each strand in the cable @xmath298 has a matching strand in the cable @xmath299 there is a canonical cobordism from @xmath242 to @xmath300 the composition of saddle point cobordisms for each pair of identically numbered strands in @xmath298 and @xmath301 ( that is , one saddle point cobordism for each dot not in @xmath274 ) , for an example see figure [ merge4 ] .
denote by @xmath302 the map of cohomology groups from @xmath303 to @xmath304 induced by this composition of saddle point cobordisms . to summarize ,
@xmath289 is the multiplication by a power of @xmath281 at certains strands of the cable @xmath305 @xmath306 is induced by annuli contraction cobordisms at `` mirrors '' of strand pairs in @xmath267 and @xmath307 and @xmath302 is induced by saddle point cobordisms of the remaining strands .
let @xmath308 @xmath309 summing over all pairings @xmath247 of @xmath20 we get a homomorphism of abelian groups from @xmath254 to @xmath310 @xmath262 is a homomorphisms of complexes .
proof is straightforward .
@xmath216 figures [ merge1]-[merge4 ] illustrate our construction for the case of @xmath267 and @xmath268 in figure [ disjoint ] . [ htb ] [ htb ] [ htb ] [ htb ]
a similar map @xmath262 can be defined for the case when a saddle point cobordism increases ( rather than decreases ) the number of components by @xmath4 we leave its construction to the reader . thus , to each reidemeister and morse move of framed colored link diagrams we can assign a map between the corresponding complexes @xmath311 we conjecture that the induced maps on cohomology of these complexes give an invariant of framed colored link cobordisms . *
categorification over a field .
* let @xmath312 be a field .
for a diagram @xmath104 of a knot @xmath23 denote by @xmath313 the complex @xmath314 and its cohomology by @xmath315 for each arrow @xmath316 we have a map @xmath317 well - defined up to overall minus sign . for each square of arrows @xmath318 the induced square of maps @xmath176
either commutes or anticommutes .
we can always make all squares of maps @xmath176 anticommute by changing signs in some of the maps @xmath234 let us call a choice of signs in maps @xmath176 _ satisfactory _ if all squares anticommute . if @xmath176 is satisfactory , the direct sum of @xmath319 over all possible @xmath184 with differential the sum of @xmath193 over all possible arrows @xmath316 forms a complex .
denote this complex by @xmath320 for any two satisfactory choices of signs @xmath321 the complexes @xmath322 and @xmath323 are isomorphic .
proofs of these two lemmas are left to the reader . from lemmas 1 and 2
we derive that the cohomology groups of @xmath324 do not depend on a satisfactory choice of signs .
denote these groups by @xmath325 yet another exercise in sign juggling shows that a reidemeister move induces an isomorphism of these cohomology groups , and implies isomorphism classes of bigraded cohomology groups @xmath326 are invariants of the framed knot @xmath28 their euler characteristic is the colored jones polynomial : @xmath327 _ remark : _ similar considerations work over @xmath328 with the further complication that we can not pass from the complex @xmath329 directly to its cohomology ( since @xmath76 is not a field ) , and will be forced to work with these complexes throughout .
our diagrams will not be commutative or anticommutative , but rather commutative or anticommutative up to chain homotopy . *
different definitions .
* there are several competing definitions of the complex whose cohomology categorify the colored jones polynomial of a link .
our maps between complexes assigned to cables go in the direction of reducing the number of strands .
we could set up these maps to go in the opposite direction . from section [ res - irr ] we know that the algebraic counterpart of the complex @xmath330 has cohomology only in degree @xmath331 taking the hint
, we could downsize our complex @xmath332 by restricting to the subgroup @xmath333 of @xmath334 the above two modifications of @xmath330 can be done independently , and give rise to the total of four complexes ( including @xmath330 ) .
we conjecture that if @xmath312 has characteristic @xmath22 these four complexes have isomorphic cohomology groups .
_ examples : _ ( a ) @xmath335 case .
we ve observed in section [ firstex ] that all four definitions give isomorphic theories over a field of characteristic different from @xmath137 \(b ) @xmath336 case over @xmath337 the differential in @xmath338 is surjective , and the composition @xmath339 is a permutation , where @xmath340 is the differential in @xmath341 it is easy to derive that all four definitions give isomorphic theories in this case . in
yet another approach , we could consider an @xmath342-action on @xmath343 induced by permutations ( braidings ) of strands of the cable @xmath344 and take the @xmath342-invariants under this action .
we expect that over @xmath345 this definition would give the same result as each the previous four .
one problem with this approach is the projectivity of the @xmath342-action , well - defined only up to sign .
in this section we categorify the reduced colored jones polynomial @xmath346 let s start with the case of a knot .
given a framed colored knot @xmath347 , the jones polynomial @xmath24 can be computed with the help of the kauffman bracket rules and the jones - wenzl projector , see @xcite .
the jones - wenzl projector is uniquely determined by graphical relations in figure [ joneswenzl ] .
[ htb ] we denote the projector by @xmath348 and by @xmath349 the projector divided by @xmath1.$ ] in this section we assume familiarity with @xcite and , in particular , with the ring @xmath350 and its indecomposable left projective modules @xmath351 for @xmath352 where @xmath353 is the set of crossingless matchings of @xmath354 points . positioning the jones - wenzl projector in the upper half of the plane we can view it as a function from @xmath353 to @xmath19,$ ] by coupling the projector to any crossingless matching , as in figure [ prmatch ] . [ htb ] the jones - wenzl projector evaluates to zero on any crossingless matching except for the one denoted @xmath355 in figure [ prmatch ] , on which it takes value @xmath1.$ ] therefore , @xmath349 is a `` delta - function '' on @xmath353 supported on @xmath356 in @xcite to any crossingless matching @xmath357 we assigned an indecomposable left projective graded @xmath350-module @xmath351 and to a tangle @xmath285 with no bottom and @xmath354 top endpoints a complex of projective left @xmath350-modules @xmath358 well - defined in the homotopy category @xmath359 of complexes of graded @xmath350-modules . to a tangle @xmath184 with no top and @xmath354 bottom endpoints we assigned a complex of projective right @xmath350-modules @xmath360
coupling @xmath285 with @xmath184 corresponds to forming the tensor product , @xmath361 for each @xmath357 there exists a right @xmath350-module , denoted @xmath362 isomorphic as a graded abelian group to @xmath76 placed in degree @xmath22 with the idempotent @xmath363 acting as the identity on @xmath362 and other minimal idempotents acting by @xmath331 if we were working over a field rather than over @xmath328 this module would have been the simple quotient of the right projective module @xmath364 while over @xmath76 the modules @xmath365 are substitutes for simple modules ( for instance , they give a basis in the grothendieck group of @xmath350-modules ) .
we have @xmath366 in particular , @xmath367 so that @xmath368 has the `` delta - function '' behaviour when coupled to projective modules . therefore , in our categorification we can interpret the quotient jones - wenzl projector @xmath349 as the right @xmath350-module @xmath369 the ring @xmath350 has infinite homological dimension , and @xmath368 does not have a finite length projective resolution .
we do not know any explicit construction of a ( necessarily infinite ) projective resolution of @xmath370 _ remark : _ for a similar categorification of the jones - wenzl projector @xmath348 we should look for a graded @xmath350-module which has a filtration with quotient modules isomorphic to @xmath371 such a module does not exist when @xmath372 we categorify the reduced colored jones polynomial of a framed knot @xmath23 with the help of @xmath370 turn @xmath23 into a tangle @xmath373 with no bottom and two top endpoints and form the @xmath0-cable @xmath374 of this tangle .
the tangle @xmath374 has no bottom and @xmath354 top endpoints .
orient it the same way we oriented cables in section [ first - cat ] .
the invariant @xmath375 is a complex of graded projective left @xmath350-modules .
let @xmath376 and define the reduced cohomology @xmath377 of @xmath23 colored by @xmath0 as the cohomology of the complex @xmath378 these cohomology groups are bigraded , @xmath379 cohomology groups @xmath377 are invariants of a framed knot @xmath28 their euler characteristic is the reduced colored jones polynomial of @xmath23 : @xmath380 _ remark : _ if @xmath142 is obtained from @xmath381 by a frame change as in figure [ curl ] , we have @xmath382 \hspace{0.1 in } \mathrm{if } \hspace{0.1 in } n=2 m , \\
\widetilde{\ch}_n(k_1 ) & \cong & \widetilde{\ch}_n(k_0 ) \{-2m(m+2)\ } [ -2m(m+1 ) ] \hspace{0.1 in } \mathrm{if } \hspace{0.1 in } n=2m+1 . \end{aligned}\ ] ] we see that @xmath383 depends on framing in a simpler way than @xmath384 _ examples : _ 1 .
when @xmath385 reduced cohomology does not depend on the knot and is isomorphic to @xmath76 placed in bidegree @xmath77 2 . for @xmath386
reduced cohomology was defined in ( * ? ? ?
* section 3 ) .
@xmath158 is a complex of @xmath387-modules , and tensoring it with the @xmath387-module @xmath76 ( where @xmath388 acts trivially ) gives us the reduced complex @xmath389 3 .
reduced cohomology of the @xmath0-colored @xmath59-framed unknot is @xmath76 placed in bidegree @xmath77 for @xmath386 the relation between @xmath390 and @xmath391 takes the form of a short exact sequence @xmath392 of complexes , giving rise to a long exact sequence in cohomology @xmath393 we do not know how to relate categorifications @xmath390 and @xmath391 for @xmath372 * categorification of the reduced colored jones polynomial of links * for simplicity , in this section we switch the base ring from @xmath76 to the @xmath102-element field @xmath337 start with a colored link @xmath394 with a distinguished component @xmath14 colored by @xmath25 turn @xmath6 into a tangle @xmath395 with no bottom and @xmath102 top endpoints by cutting a segment out of @xmath396 taking the @xmath0-cable of @xmath395 at the component @xmath14 gives us a tangle , denoted @xmath397 with no bottom and @xmath354 top endpoints .
apply the construction of section [ first - cat ] to all components of @xmath398 other than the @xmath0 components coming from @xmath396 the result is a complex of projective left @xmath350-modules which we denote @xmath399 let @xmath400 ( where @xmath401 is the right @xmath402-module @xmath403 ) , and define the reduced cohomology @xmath404 as the cohomology of the complex @xmath405 isomorphism classes of cohomology groups @xmath404 are invariants of a framed colored link @xmath6 with a distinguished component @xmath396 their euler characteristic is the reduced colored jones polynomial @xmath346
@xmath406 \(a ) for a clean definition of cohomology groups @xmath407 it is necessary to understand the sign ambiguity of our cohomology theory . this ambiguity forced us into a clumsy definition of @xmath408 for a field @xmath312 of characteristic other than @xmath120 see section [ first - cat ] .
no matter how we define @xmath409 taking care of signs is a prerequisite for extending these homology groups to an invariant of ( framed and colored ) link cobordisms .
\(b ) cohomology theory described in section [ first - cat ] depends nontrivially on the framing , and is expected to give rise to invariants of _ framed _ cobordisms between framed links . to check the invariance
, one needs a list of movie moves for framed cobordisms .
\(c ) give a clean definition of cohomology groups @xmath409 preferably over @xmath328 and extend this construction to a functor from the category of ( colored , framed , decorated ) link cobordisms to the category of bigraded abelian groups . | the colored jones polynomial of links has two natural normalizations : one in which the @xmath0-colored unknot evaluates to @xmath1,$ ] the quantum dimension of the @xmath2-dimensional irreducible representation of quantum @xmath3 and the other in which it evaluates to @xmath4 for each normalization we construct a bigraded cohomology theory of links with the colored jones polynomial as the euler characteristic .
_ _ # 1#2#3#4 \
{ ll # 1 & # 2 + # 3 & # 4 .
# 1#2 ( c # 1 + # 2 ) # 1 |
the minor planet ( 87 ) sylvia is a main belt asteroid discovered in 1866 by @xcite . in the 1990s ,
frequency analysis of photometric observations hinted that this asteroid could be binary @xcite .
its first satellite ( s/2001 ( 87 ) 1 , known as _ romulus _ ) was discovered in february 2001 by @xcite using the keck ii telescope atop hawaii s mauna kea .
three years later , @xcite announced the discovery of a second companion ( s/2004 ( 87 ) 1 , known as _ remus _ ) , using the european southern observatory s very large telescope ( vlt ) .
sylvia became the first known triple asteroidal system .
since then , eight others have been discovered and studied @xcite .
asteroid ( 87 ) sylvia is the largest member of a collisional family born , at least , several hundreds of million years ago , more probably between 1 and 3.8 gyr .
the age of this family , for which more than 80 members have been identified among current census of asteroids , is commensurable with the evolutionary timescales of sylvia s satellite system @xcite .
various authors estimate that the system is dynamically very stable over a large timescale ( at least one million years , see @xcite ) , the satellites being in a deeply stable zone , surrounded by both fast and secular chaotic regions due to mean - motion and evection resonances @xcite .
observations led to the determination of the dynamical and physical properties of the system .
asteroid ( 87 ) sylvia is classified as a x - type asteroid @xcite located in the outer main belt ( @xmath6 au , @xmath7 , @xmath8 ) , within the large cybele - zone , with a volume - equivalent diameter estimated to @xmath9 km , a relatively low density of @xmath10 , and a large macro - porosity estimated to @xmath11% ( @xcite , and references therein ) .
the two moons , remus and romulus , with a diameter respectively estimated to @xmath127 km and @xmath1218 km ( from photometry measurements , * ? ? ? * ) or 912 km and 516 km ( derived by * ? ? ? * as a free parameter of their dynamical model ) , orbit at a distance of @xmath12700 km and @xmath121350 km from the primary . finally , @xcite estimates dynamically that sylvia is oblate with a @xmath13 value in the range 0.09850.1 .
we report here on new results on the dynamical and physical properties of sylvia s system based on the analysis of adaptive - optics imaging , light - curves and stellar occultation data ( section [ data ] ) .
we improve sylvia s 3-d shape model and estimate its overall size ( section [ sylvia ] ) .
we estimate the shape and size of the outer satellite romulus from the analysis of the latest observed stellar occultation ( section [ romulus ] ) .
we improve the determination of orbital parameters for the two satellites , we estimate the mass and density of sylvia , and we examine its quadrupole term @xmath13 ( section [ orbits ] ) .
finally , we discuss the surprising elongated shape of romulus revealed by this stellar occultation .
we gathered in the vobad database @xcite all observations , acquired by our group or already published , from february 2001 to december 2012 recorded in the near - infrared with adaptive - optics ( ao ) systems available on large ground - based telescopes .
we use the eso very large telescope naco imaging camera @xcite and sinfoni spectro - imaging camera @xcite , the gemini north altair ao system @xcite with its camera niri @xcite , and the nirc2 camera on the keck ii telescope @xcite .
the ao frames were recorded in broad band filters ( j , h , or k , from 1 to 2.5 @xmath14 m ) and were all processed in a similar manner . the basic data processing ( sky subtraction , bad - pixel removal , and flat - field correction ) applied on all these raw data was performed using the recommended eclipse data reduction package @xcite .
successive frames taken over a time span of less than 6 min , were combined into one single average image after applying an accurate shift - and - add process through the jitter pipeline offered in the same package .
data processing with this software on such high signal - to - noise ratio data ( @xmath151000 ) is relatively straightforward . since these data respect the shannon s theorem , it is possible to retrieve completely the continuous signal from the knowledge of enough samples . after re - sampling
each image to 1/10th of the pixel level , the centroid position on each frame can be accurately measured by a gaussian fit .
the final image is obtained by stacking the set of frames with individual shifts determined from cross - correlation .
once processed , individual images reveal the resolved shape of the primary ( angular size @xmath16 ) , and , sometimes , the unresolved image of the satellites appears .
we used a dedicated algorithm @xcite to extract the contour of the primary and to determine its photocenter , from which we measured the astrometric positions of the satellites by fitting a moffat - gauss source profile .
figure [ fig - ao ] displays sylvia s system as seen by vlt / naco and keck / nirc2 instruments after the processing has been applied .
table [ tbl - observingconditions ] provides the observing condition at each epoch of image acquisition .
tables [ tbl - obsremus ] and [ tbl - obsromulus ] summarize all the astrometric measurements used to fit the orbits of the two satellites of sylvia .
the accuracy on the observing time is the result of the computed mean time of the jittered images , typically 0.2s to 1s depending on the observatory where the data were recorded .
@xcite reported the confirmation of the presence of romulus on hubble space telescope ( hst ) images collected on february 23 , 2001 with the wfpc2 instrument through various filters .
we retrieved from the hst archive the three unsaturated observations taken through the f439 filter with an individual exposure time of 3s .
these observations were re - processed using our own pipeline reproducing the hst / wfpc2 cookbook method . the resulting image , shown in fig .
[ fig - ao ] , confirmed the detection of romulus at a position very close to the one reported by @xcite .
the second satellite , remus , is also visible and detected closer to the primary at a distance of 0.34@xmath17 .
the positions of the satellite were derived using the same moffat - gauss profile fit than for our ao observations .
interestingly , even though @xcite reported the triple nature of ( 87 ) sylvia from observations taken in august 2004 and onward , this february 2001 hst observation was in fact the first detection of the third component of the system .
we included these astrometric positions for romulus and remus in our analysis , which is particularly useful for remus since it increases the observational time span by 1264 days ( @xmath12931 revolutions ) .
we used 40 light - curves observed from 1978 to 1989 published by @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
the data were compiled by @xcite in the uppsala asteroid photometric catalog , now available through internet ( http://asteroid.astro.helsinki.fi/apc[apc ] web site ) . to this set of dense light - curves we added sparse photometry from us naval observatory ( iau code 689 ) , roque de los muchachos observatory , la palma ( 950 ) , and catalina sky survey observatory ( 703 ) .
see the works by @xcite and @xcite for details on sparse photometry .
the observation of a stellar occultation consists in recording the time of immersion and emersion of the star in front of which the asteroid passes , as seen by geographically distributed observers along the occultation path .
each observed occultation point is then projected onto a common plane that passes through the center of the earth , and lies perpendicular to the direction of the star as seen from the occulting body . assuming that the relative velocity of the body with respect to the observer is well estimated by the ephemeris , which is a soft assumption , especially for numbered asteroids
, one can transform the timings into lengths , and then evaluate the size of the occulting body .
we get several segments - the chords - which are directly proportional to the size of the cross - section of the body as seen by observers . with a sufficient number of chords ,
the silhouette of the body is drawn , and can yield a strong constraint on its 2-d profile in the occultation plane .
the observation of a stellar occultation by an asteroid is one of the few methods which can yield the size and the overall shape of the asteroid without hypothesis on its physical nature .
if several independent events of the same occulting body are collected then a 3-d model of the asteroid can even be determined @xcite .
however , due to the low number of well - covered events for a given asteroid , full 3-d reconstruction based on stellar occultations only will always concern a small sample of asteroids .
stellar occultations have however proved to be useful to scale the convex shape models of asteroids derived by light - curve inversion @xcite .
four stellar occultations by sylvia have been reported in the past 30 years , but only a total of four chords ( three for an event , one for another , none for the two others ) have been collected @xcite .
these occultation data are therefore useless to scale sylvia s shape model @xcite .
the stellar occultation by sylvia successfully observed in early 2013 is thus the first opportunity to do so . on january 6 , 2013
, about 50 european observers were mobilized to observe the occultation of the tycho-2 1856 - 00745 - 1 star by ( 87 ) sylvia @xcite . among them , 19 observers have recorded a negative event ( i.e. , no disappearance of the star ) , and 13 observers have reported a successful observation of the event , providing 16 chords including 4 of the occultation by romulus .
the bad weather forecast on western europe this night prevented other observers to record the event .
figure [ fig - occhords ] shows the result of the observation of this stellar occultation .
table [ tbl - timings ] presents the timings of the event recorded by observers ( published on http://euraster.net/results/2013/index.html#0106-87[euraster ] web site ) , and table [ tbl - observers ] lists the observers and the geodetic coordinates of the observing sites .
stellar occultations by asteroids are usually observed by a group of observers who use different acquisition and timing devices . as a consequence ,
the accuracy on the measurements differs from one observer to another , and sometimes measurements can disagree owing to systematics in the calibration of the absolute timing reference .
a typical example is a chord which is clearly shifted with respect to other chords nearby .
the latter can then be used to estimate the offset to apply on the chord to restore its timings .
however , it can be tricky to shift the timings of chords , mainly because no evident rule can be found . in such cases
it is better to decrease the weight of uncertain chords with respect to accurate ones . for the january 6 , 2013 stellar occultation by sylvia , we are confident in the absolute timings of the chords as most timing references were insured by gps or atomic clock time servers ( see tab .
[ tbl - observers ] ) .
we considered only two chords as discrepant ( see column `` offset '' of tab .
[ tbl - timings ] ) .
the first shows a clear lateness of a few seconds with respect to the three chords immediately next to it ( one north , two south ) .
the offset we apply ( -3.8s ) sets the chord back in alignment with the three others .
the second discrepant chord shows a slight lead on a nearby chord for which the observer can assert the time of the disappearance .
the offset we apply ( + 0.5s ) shift the chord forward in alignment with the latter . as we empirically changed the timings of these two chords , we down - weighted their timings to one thousandth of the weight affected to other chords .
at the mean time of the occultation , january 6 , 2013 01:37:06.46 utc , the computed geocentric orbital velocity of sylvia is 17.1 km / s . with timing uncertainties lying in the range 0.031s ( with a mean value of @xmath18s ) , we can expect to measure lengths with a mean accuracy of 4 km ( @xmath121.5% of sylvia s size ) , which corresponds to @xmath122 mas at the geocentric distance of sylvia .
a convex shape model of sylvia was previously derived by @xcite .
they used 32 light - curves observed in 8 apparitions spanned over 1978 - 1989 , which allowed them to derive a unique model with ecj2000 pole direction ( 71 , + 66 ) .
as was later shown by @xcite , this model agreed well with ao observations .
it also agrees with the occultation data presented here .
to further improve the fit and to include all data types into the modeling , we used the koala algorithm @xcite .
it can simultaneously fit optical light - curves , occultation data and image profiles to give a best - fit 3-d shape model represented by a non - convex polyhedron @xcite .
contrary to standard light - curve inversion , where light - curves are fitted by a convex shape model , the best - fit solution is harder to define in case of multiple data sources .
different types of data have to be weighted to compose the total @xmath19 as a measure of the goodness of fit @xcite .
moreover , given the low number of 2-d profiles available , the non - convex shape solution had to be regularized to suppress unrealistic small - scale fluctuations of the shape .
the shape was approximated by a polyhedron described by a spherical harmonics series of the order and degree five or six .
this resolution was sufficient to model details of the occultation profile , and still low enough to suppress artificial details that often appeared with higher resolution .
the koala algorithm minimizes the total @xmath20 that composes the individual contributions from light - curves ( lc ) , adaptive optics profiles ( ao ) , and occultation chords ( occ ) .
adaptive optics and occultation data are weighted with respect to the light - curves with parameters @xmath21 and @xmath22 , respectively .
the optimum values of these weights ( so called maximum compatibility weights ) can be obtained following the approach of @xcite . in our case ,
the shape of the @xmath23 surface was ` shallow ' , thus the optimum weights were not well defined
an order of magnitude range of @xmath21 and @xmath22 was possible .
we varied weights of ao and occultation data with respect to light - curves to see how the models evolved , then we selected only solutions that gave acceptable fits to the data .
thus , instead of one best solution , we obtained a set of slightly different models that all fit the data well , each with a slightly different spin , shape , and size .
one of these models is shown in fig .
[ fig - onebestmodel ] , and the fits to the light - curves , image and occultation profiles are shown in fig .
[ fig - colfits ] .
the discrepancy between one ao image contour and the model silhouette comes from an over - estimation of the ao contour due to poor seeing conditions at the time of this observations .
the root mean square ( rms ) residual of the occultation fit is 2.6 km , which is below the accuracy provided by the timings ( see [ occdata ] ) . to further estimate the uncertainty of our model
, we created many clones of the original data set and processed them the same way as the original data .
for light - curves , we re - sampled the set with random sampling and replacement so each new clone consisted of the same light - curves , but some were missing and some were included more times .
this approach is similar to the bootstrap method @xcite , the difference is that we re - sampled the whole light - curves , not individual measurements , because points within one light - curve can not be treated as fully independent measurements . by this approach
, we can roughly sample the range of possible solutions .
for occultation data , we randomly changed the timings using a normal distribution law , with reported uncertainties as standard deviations .
because of the low weight of the ao data in the process , we left them unchanged .
the distribution of shapes and spin axis directions then represented the variance of the model given the data and served for a realistic estimation of the model uncertainties .
the width of @xmath24 error intervals reported in table [ tbl - physparams ] corresponds to the maximum spread of the relevant parameters across our sample of bootstrap models obtained with different @xmath21 and @xmath22 weights .
figure [ fig - rmspole ] displays the rms residuals between synthetic and observed light - curves as function of the spin - vector coordinates ( grid interval of 2on the ecj2000 celestial sphere ) . for one of the formally best solutions ( based on the total @xmath19 ) ,
the direction of the spin axis ( 70 , + 69 ) is almost the same as for the convex model of @xcite , and the 3-d shapes are similar overall .
the standard deviation of the pole direction is about 3 in ecliptic longitude and 1 in ecliptic latitude .
the volume - equivalent diameter for the best model is @xmath3= 273 km , and @xmath13 = 0.024 .
the basic physical parameters are listed in table [ tbl - physparams ] .
we also used different resolutions of the model ( the degree and order of spherical harmonic expansion used for the shape parametrization ) and different weightings between the data to assert the stability of the solution .
the plausible shape models cover a range of equivalent diameters from 266 to 280 km and @xmath13 spans the range 0.0150.040 .
using our 3-d shape model , we estimated for each image the offset between the center of mass and the photocenter , assuming an uniform density distribution .
they span the intervals [ -2.3 , + 1.9 ] mas in @xmath25 , and [ -0.4 , + 1.8 ] mas in @xmath26 , and are , at least , five time smaller than the uncertainties of the astrometric measurements ( see tables [ tbl - obsremus ] and [ tbl - obsromulus ] ) .
we have not taken into account these offsets in the astrometric measurements used to fit orbits .
the analysis of the chords of romulus give us the opportunity to determine its size and , for the first time , to estimate its shape .
the distribution of the 4 observed chords is not sufficient to model without ambiguity the profile of the satellite , but they are sufficiently well distributed to estimate its size and biaxial dimensions .
the parameters of the ellipsoidal profile ( coordinates of the center @xmath27 , axis radii @xmath28 , and position angle from north of the major axis @xmath29 ) are calculated from the set of conics which best fit the @xmath30 points of coordinates @xmath31 , which correspond to the coordinates of the extremity of the chords as measured by the observers .
we solve the system of linear equations by the linear least - squares method of the singular value decomposition method @xcite .
each fitted points is assigned initially a standard deviation calculated as the quadratic mean of the observational uncertainties .
even if our system is overdetermined ( @xmath32 equations for 5 parameters ) , the solution can be singular , or numerically very close to singular . in that case , we reduce the system to 3 parameters by fixing the coordinates of the conic center ( @xmath27 ) .
then we define a grid of values with a given step size , and we solve the equations for each point of the grid , searching for the conics which have @xmath33 , the 3@xmath34 confidence region for the normal distribution with 3 degrees of freedom @xcite . among the observers ,
two could have observed the occultation by romulus , namely m. bretton and v. fristot ( c.f .
[ tbl - timings ] and [ tbl - observers ] ) .
the first one made a naked eye observation and stop too early to observe romulus event .
the analysis of the video acquired by the second confirms that no occultation of romulus occurred along his path .
this result sets a strong upper limit on the size of romulus , which allows us to reject all the fitted conics which intersect this chord . considering a range of 20 km wide for the coordinates of the conic center , and a grid step size of @xmath35 km ,
the best - fit conic profile of romulus is ( fig .
[ fig - romulusbestmodel ] ) : @xmath36 with uncertainties at 3@xmath34 .
it implies a mean axis ratio @xmath37 , and a surface - area equivalent diameter @xmath38 km .
this result relies on the reliability of the timings of the two northern chords of romulus .
even if p. tanga failed to report accurate absolute timings , he was able to estimate them using the computer clock ( synchronized by ntp ) used to control the camera , and to measure precisely the duration of events and the delay between the two events ( see tab .
[ tbl - timings ] ) . using the 3-d shape model of sylvia projected into the occultation plane
, we can tie the time of the first disappearance of the star observed by p. tanga to other timings with an accuracy smaller than 0.2s . in the case we shift tanga s chords by + 0.2s
, the axis ratio of the best - fit profile of romulus decreases to 2.2 , with @xmath39 km , @xmath40 km , @xmath41 km , and equivalent uncertainties .
orbital solutions for sylvia s satellites have been recently published by @xcite , and previously by @xcite at the occasion of the discovery of the second satellite .
we take the opportunity of the successful observation of a stellar occultation by romulus to test the reliability of its orbital solution , used to predict the event @xcite , and to improve the orbital parameters of the two satellites .
we used our genetic - based algorithm _
genoid - kepler _
@xcite to predict the position of sylvia s satellites for the january 6 , 2013 stellar occultation .
the dynamics used to model the orbit of the satellites was a pure 2-body keplerian problem ( no mutual perturbation between the two satellites was applied ) . for remus , we used 25 astrometric positions ( acquired in 2001 , 2004 , 2010 and 2011 ) spanning 3948.7 days or 2910.6 revolutions . for romulus , we used 65 astrometric positions ( acquired in 2001 , 2004 , 2005 , 2006 , 2010 and 2011 ) spanning 3953.8 days or 1317.9 revolutions .
tables [ tbl - observingconditions ] , [ tbl - obsremus ] and [ tbl - obsromulus ] detail the observing condition and the astrometric positions used for this work .
the best solutions obtained give orbits with a rms of 12.1 mas for remus , and 11.2 mas for romulus .
the successful observation of the occultation by romulus highlights the reliability of our dynamical model .
figure [ fig - romulusbestmodel ] shows the goodness of the prediction of romulus position with respect to sylvia s center of mass ( as defined by the 3-d model ) .
the offset between the observed position of romulus ( e.g. , defined by the center of the fitted profile ) and its predicted position is @xmath42 km , @xmath43 km ( @xmath25 and @xmath26 are positive toward east and north ) .
this predicted position is located at 13.5 km from the observed position . in the occultation plane ,
the distance between the fitted profiles of sylvia and romulus is 1154.6 km .
the geocentric orbital velocity of sylvia being 17.1 km / s , the occultation by romulus occurred 67.6 s later than sylvia event . at this epoch ,
january 6 , 2013 01:38:14.02 utc , we infer the astrometric position of romulus relatively to sylvia s center of mass : @xmath44 mas .
we added this astrometric position into _ genoid - kepler _ input dataset , as well as an astrometric position acquired in 2010 with vlt / sinfoni @xcite .
it increases the time span of data from 3953.8 to 4338.7 days or 1446.2 revolutions .
this does not change significantly the dynamical solution of romulus ( rms = 11.2 mas ) .
table [ tbl - dynamicalparam ] presents the orbital parameters of sylvia s satellites obtained with _ genoid - kepler _ , and figure [ fig - omc - sigma ] displays the residuals ( the difference between the observed @xmath45 and computed @xmath46 positions of the satellite and the primary ) normalized by the positional uncertainty @xmath34 .
the orbital solutions indicate that the satellites follow a quasi circular orbit , nearly coplanar ( very low mutual inclination , lesser than @xmath47 ) , and nearly aligned with the primary equatorial plane ( by @xmath124 ) .
the average residuals are @xmath48 mas , @xmath49 mas for remus , and @xmath50 mas , @xmath51 mas for romulus ( uncertainties at 3-@xmath34 ) , matching the level of accuracy of the observations . at the epoch of the occultation , the reliability of the model can be estimated by the difference between the observed and computed positions of romulus , which are @xmath52 mas and @xmath53 mas , or @xmath54 km and @xmath55 km in the occultation plane .
we inserted these orbital solutions in the web service http://vo.imcce.fr/webservices/miriade[miriade ] of our virtual observatory solar system portal , allowing everyone to compute the ephemeris of sylvia s satellites .
the mass of sylvia can be derived from the best - fit keplerian orbit of the satellites ( see tab .
[ tbl - dynamicalparam ] ) . assuming the gaussian gravitational constant and the solar mass
, one can easily derive , independently , the mass of sylvia from the semi - major axis and the orbital period of each satellite ( neglecting its own mass ) .
we find @xmath56 kg from remus , and @xmath57 kg from romulus , considering the 1@xmath34 uncertainties of the parameters .
it is interesting to note that the two masses are consistent within their uncertainties .
the mass derived from romulus orbit is also very similar , by @xmath120.5% , to that found by @xcite , although it has been obtained by the latter by the fit of a fully dynamical three - body model . from the mass of sylvia and its volume , defined by its 3-d shape model
, we infer its density .
we find 1.29 and 1.38 g.@xmath58 from remus and romulus data respectively , with an uncertainty of @xmath59 g.@xmath58 , taking into account the 1@xmath34 uncertainties of the parameters .
the resulting density of [email protected] g.@xmath58 is slightly lower , although consistent , with the average density of x - type asteroids of [email protected] g.@xmath58 @xcite .
the composition and meteorite analogs of x - types is still debated , and proposed analogs encompass cv and co carbonaceous chondrites , enstatite chondrites , aubrites , mesosiderites , and even iron , stony - iron meteorites @xcite , which density ranges from 2.8 to 7.7 g.@xmath58 .
although we do not have information on the composition of sylvia , its density , lower than that of its most - likely components , indicates the presence of voids and/or pockets of very low density material ( e.g. , ices ) .
the macroporosity we derive here ranges from 52% to 82% depending on the density of the analog meteorite .
we used our _ genoid - anis _
algorithm @xcite to search for a possible influence of an irregular mass distribution , considering the amount of voids determined above .
we initialized the problem with @xmath61 , the value estimated from the 3-d shape model ( see sec .
[ sylvia ] ) .
we search for orbital solutions in the space @xmath62 , all other parameters being free .
the results are shown in fig .
[ fig - j2 ] . assuming that the directions of the orbital poles of the satellites must be nearly aligned with the primary pole of rotation ( hypothesis justified by the low inclination of orbital planes , see sec .
[ newsoluce ] ) , we discard all solutions for which coordinates of orbital poles are outside the 3@xmath34 confidence interval of the primary spin - vector coordinates ( see sec .
[ sylvia ] and fig .
[ fig - rmspole ] ) .
we find that the best candidate solutions are obtained for @xmath63 .
it means that no significant precession effect of the apsidal and nodal nodes is detected , and that the purely keplerian orbit is enough to fit the available data to their accuracy .
@xcite showed that sylvia s system is not stable unless the primary has a minimal amount of oblateness , at least 0.1% of the assumed primary @xmath5 of 0.17 reported by @xcite .
this oblateness provides a faster stabilizing effect on the satellites orbits than other gravitational perturbations ( e.g. , sun , jupiter , see also @xcite ) .
our @xmath64 result does not contradict this statement , insofar as we analyze observational data and do not study the long term stability of the system .
it shows that the estimation of the quadrupole term @xmath13 of the gravitational potential of sylvia is not obvious , even impossible , given the available observational data , especially since a keplerian model provides the best - fit orbital solution .
the discrepancy between our result and those of @xcite ( who found a primary @xmath5 of 0.1 ) , or those of @xcite , reveals that the estimation of the @xmath13 term is strongly correlated with the orbit - fitting method . in all cases ,
very similar orbital solutions are obtained , and the same conclusion applies : sylvia s system is ancient and in a very stable state .
the difference in @xmath5 from our dynamical analysis and as derived from the 3-d shape model implies that the assumption of an homogeneous mass distribution is not valid .
the @xmath64 determined dynamically indicates a more concentrated mass distribution than the 3-d shape suggests , so that the primary could be differentiated with a dense core .
the same result emerged from the analysis of the orbit of the 28 km - diameter satellite of ( 22 ) kalliope , a 166 km - diameter m - type asteroid .
@xcite suggested a differentiated internal structure for the primary of this binary system .
recent developments ( e.g. , * ? ? ?
* ) in thermal modeling of small solar system bodies showed that large enough ( radius @xmath15 7 km ) asteroids that accreted shortly after the calcium - aluminum inclusion formation ( @xmath65 2.5 myr ) , when @xmath66al was still abundant , could have been molten internally . as they were cooling off , their internal composition became partially differentiated .
those models showed that these asteroids remained covered with a thick ( up to tens of kilometers ) un - melted relict crust ( see a review on this mechanism and a thorough discussion in * ? ? ?
although we can not conclude whether or not the internal structure of ( 87 ) sylvia is differentiated , the very low @xmath5 , perturbing the satellite s orbit , hints toward a dense core embedded into an irregularly shaped material . the lower density of the surrounding material may be related to composition or structure , e.g. , a large macro - porosity generated by fractures .
the shape and size of romulus was estimated in section [ romulus ] from the chords of an occultation , leading to a surface - area equivalent diameter @xmath0 km ( h = 12.112.3 , assuming the same albedo as the primary of 0.0435 , @xcite ) , and an axis ratio @xmath67 ( assuming an ellipsoid ) .
this is the second time the shape and size of a satellite of asteroid is measured using this technique .
an occultation by linus , satellite of the asteroid ( 22 ) kalliope , gave a similar opportunity @xcite , but the chords were too close to each other and the shape could only be approximated by a circular fit . in the case of romulus , the positive and negative chords are sufficiently spread ( fig .
[ fig - romulusbestmodel ] ) to give a meaningful constraint on its shape .
the satellite appears extremely elongated , with the main axis of the silhouette ellipse oriented at @xmath68 from the primary .
we searched in the _ asteroid lightcurve database _
@xcite containing the light - curves of 6160 small solar system bodies ( including neas , tnos , main - belt and a few comets ) those with an amplitude greater than 1.9 ( corresponding to a size ratio @xmath69 2.4 ) and found 15 asteroids , including 12 neas and 3 main - belt asteroids ( fig .
[ fig - lcdb ] ) .
they are all significantly smaller than romulus with a diameter varying from 20 m to 7.5 km .
the largest , ( 44530 ) horakova , has a light - curve poorly constrained with a period @xmath70 h and a maximum light - curve amplitude @xmath71 , suggesting that it could be a tidally locked binary .
asteroid ( 1620 ) geographos ( @xmath72 km , @xmath73 ) is smaller but shows the same light - curve amplitude as romulus .
moreover its shape is well - defined thanks to delay - doppler radar observations @xcite .
shape reconstruction from radar gives @xmath74 , similar to romulus .
@xcite has shown by numerical simulations that tidal disruptions during a close encounter with a planet could produce the elongated shape of geographos .
similarly , because romulus is relatively close to the primary ( @xmath1210 @xmath75 sylvia s radius ) , and could be a rubble - pile satellite which formed from fragments of a catastrophic collision on the parent body , its elongated shape may result of the tidal forces from the elongated and spinning primary .
another possibility is that romulus is a bilobed satellite , relic of its formation by a low - relative speed encounter of two @xmath1210 km fragments of the parent body and captured in the gravitational field of the primary .
an accurate modeling of the tidal evolution of the orbit and shape of the satellite could shed light on the internal stress and cohesive forces of the satellite .
this work demonstrates once again that the combination of adaptive - optics observations with light - curve photometric observations and stellar occultations is a powerful way to study multiple asteroid system .
similarly to our work on the triple asteroid ( 93 ) minerva @xcite , we derive here the 3-d shape and size of sylvia s primary and constrain its internal structure . the successful observation of the occultation by romulus , the outer satellite of the system , provides the first well - constrained estimate of the shape of an asteroid s moon which has an extreme elongation , likely due to the tides due to the primary .
a follow - up of the orbits of sylvia s moons based on additional ao observations , recorded with present ao systems and on the next generation of adaptive optics currently being designed for the keck telescope @xcite or set up on new telescopes ( e.g. , lbtao , see @xcite ) , will help to confirm the absence of precession indicative of a heterogeneous interior for this asteroid
. the analysis of the occultation chords recorded on january 6 , 2013 does not reveal the presence of remus , the inner satellite , because of a poor coverage of the northern part of the occultation path ( fig .
[ fig - occhords ] ) , where remus was predicted to be located .
no observer reported occultation events due to the presence of smaller , yet unknown , satellites around sylvia s primary .
from the timing accuracy of recorded light - curves ( tab .
[ tbl - timings ] ) , we estimate that observers could have detected other satellites around sylvia with a diameter larger than 2 km along the occultation chords ( fig .
[ fig - occhords ] ) . for comparison ,
the upper limit of detection for the ao observations varies with the distance to the primary and the quality of the ao correction ; typically a satellite larger than @xmath125 km in diameter could have been detected if located at remus distance ( @xmath76 ) .
this is a clear illustration of the usefulness of occultation that help determining the multiplicity of an asteroid , even if it is far , too faint , and/or accompanied with a small moon undetectable by ao systems . as an example
, the observation of an occultation by the jupiter - trojan asteroid ( 911 ) agamemnon in 2012 showed a deep brief secondary event that is likely due to a satellite of about 5 km @xcite .
additional occultations involving ( 87 ) sylvia are predicted in the near future .
table [ tbl - futurpred ] lists a selection of events scheduled for the next 10 years , and their locations . in the future , we will continue to monitor the orbits of the satellites and deliver predictions of the moons paths at the time of occultations .
we are grateful to all the observers for their efforts and for providing their observations ( tab .
[ tbl - observers ] ) .
we appreciate the help of s. preston ( iota ) who made available the prediction of the event , to h. harris ( unso / flagstaff ) and b. owen ( tmo ) for providing last minute astrometric observations of the asteroid , to p. descamps and _ les makes _ observatory ( la runion island ) for the astrometric measurements of the occulted star , to e. frappa ( euraster ) , j. lecacheux ( lesia / observatoire de paris ) et al . to have organized the observational campaign and collected the observer reports .
the work of f. marchis was supported by the nasa grant nnx11ad62 g .
the work of j. urech was supported by the grants gacr 209/10/0537 and p209/12/0229 of the czech science foundation .
we thank the c2pu team who gave us access to the data collected by the _ omicron _
telescope of the _ centre pdagogique plante et univers _
( observatoire de la cte dazur , calern site ) , as well as the tarot observatory team who made available their robotic telescopes to record occultation events . 2001 - 02 - 19 08:39:46.94 & 12.41 & 2.791595 & 3.755543 & 3.95 & 164.84 & 1.123 & keck / nirc2 & a + 2001 - 02 - 19 10:03:57.89 & 12.41 & 2.791677 & 3.755554 & 3.96 & 164.81 & 1.014 & keck / nirc2 & a + 2001 - 02 - 19 10:12:43.20 & 12.41 & 2.791686 & 3.755555 & 3.96 & 164.81 & 1.010 & keck / nirc2 & a + 2001 - 02 - 20 09:39:27.36 & 12.41 & 2.793272 & 3.755729 & 4.08 & 164.33 & 1.027 & keck / nirc2 & a + 2001 - 02 - 20 09:46:55.78 & 12.41 & 2.793281 & 3.755730 & 4.08 & 164.33 & 1.021 & keck / nirc2 & a + 2001 - 02 - 23 13:40:12.58 & 12.45 & 2.800411 & 3.756284 & 4.58 & 162.37 & - & hst / wfpc2 & b + 2004 - 07 - 25 10:22:17.76 & 12.14 & 2.410849 & 3.218312 & 12.82 & 135.39 & 1.205 & vlt / naco & c + 2004 - 08 - 10 07:17:16.22 & 11.89 & 2.293127 & 3.217288 & 8.99 & 150.30 & 1.008 & vlt / naco & c + 2004 - 08 - 29 07:18:19.30 & 11.64 & 2.230743 & 3.216817 & 5.05 & 163.73 & 1.115 & vlt / naco & c + 2004 - 08 - 29
07:26:17.95 & 11.64 & 2.230739 & 3.216817 & 5.05 & 163.73 & 1.133 & vlt / naco & c + 2004 - 08 - 29 08:46:08.83 & 11.64 & 2.230705 & 3.216817 & 5.04 & 163.75 & 1.445 & vlt / naco & c + 2004 - 09 - 01 05:54:55.01 & 11.63 & 2.229630 & 3.216818 & 4.89 & 164.26 & 1.017 & vlt / naco & c + 2004 - 09 - 01 06:03:55.01 & 11.63 & 2.229631 & 3.216818 & 4.89 & 164.26 & 1.024 & vlt / naco & c + 2004 - 09 - 01 08:25:31.58 & 11.63 & 2.229646 & 3.216818 & 4.89 & 164.27 & 1.406 & vlt / naco & c + 2004 - 09 - 03 07:09:06.34 & 11.63 & 2.230255 & 3.216830 & 4.90 & 164.22 & 1.148 & vlt / naco & c + 2004 - 09 - 04 08:41:35.81 & 11.64 & 2.231039 & 3.216840 & 4.94 & 164.07 & 1.601 & vlt / naco & c + 2004 - 09 - 05 04:09:27.94 & 11.64 & 2.231820 & 3.216849 & 4.99 & 163.89 & 1.015 & vlt / naco & c + 2004 - 09 - 05 08:15:20.74 & 11.64 & 2.232024 & 3.216852 & 5.01 & 163.85 & 1.452 & vlt / naco & c + 2004 - 09 - 06 03:50:04.13 & 11.65 & 2.233033 & 3.216863 & 5.08 & 163.61 & 1.027 & vlt / naco & c + 2004 - 09 - 07 02:36:23.04 & 11.65 & 2.234453 & 3.216878 & 5.18 & 163.28 & 1.142 & vlt / naco & c + 2004 - 09 - 07 09:24:57.31 & 11.65 & 2.234943 & 3.216883 & 5.21 & 163.17 & 2.283 & vlt / naco & c + 2004 - 09 - 08 06:54:52.01 & 11.66 & 2.236557 & 3.216899 & 5.33 & 162.78 & 1.171 & vlt / naco & c + 2004 - 09 - 11 04:51:29.95 & 11.69 & 2.243369 & 3.216966 & 5.80 & 161.20 & 1.006 & vlt / naco & c + 2004 - 09 - 13 03:25:24.10 & 11.72 & 2.249181 & 3.217022 & 6.17 & 159.92 & 1.020 & vlt / naco & c + 2004 - 09 - 13 05:30:10.66 & 11.72 & 2.249465 & 3.217024 & 6.19 & 159.86 & 1.047 & vlt / naco & c + 2004 - 09 - 14 03:49:52.03 & 11.73 & 2.252630 & 3.217054 & 6.39 & 159.19 & 1.003 & vlt / naco & c + 2004 - 09 - 14 06:36:31.10 & 11.73 & 2.253047 & 3.217058 & 6.41 & 159.10 & 1.199 & vlt / naco & c + 2004 - 09 - 14 06:44:34.08 & 11.73 & 2.253068 & 3.217058 & 6.41 & 159.10 & 1.225 & vlt / naco & c + 2004 - 09 - 15 05:08:21.70 & 11.75 & 2.256499 & 3.217090 & 6.62 & 158.39 & 1.032 & vlt / naco & c + 2004 - 09 - 15 05:16:22.94 & 11.75 & 2.256520 & 3.217090 & 6.62 & 158.39 & 1.042 & vlt / naco & c + 2004 - 10 - 19 00:27:06.91 & 12.30 & 2.518403 & 3.219585 & 14.37 & 126.79 & 1.038 & vlt / naco & c + 2004 - 10 - 20 00:08:28.03 & 12.31 & 2.529405 & 3.219697 & 14.54 & 125.85 & 1.057 & vlt / naco & c + 2004 - 10 - 25 06:28:30.72 & 12.39 & 2.590441 & 3.220330 & 15.39 & 120.89 & 1.359 & keck / nirc2 & c + 2004 - 11 - 02 07:34:37.34 & 12.51 & 2.690426 & 3.221417 & 16.43 & 113.48 & 1.428 & gemini / niri & c + 2005 - 11 - 01 13:18:57.31 & 11.99 & 2.442235 & 3.394667 & 5.81 & 159.75 & 1.119 & gemini / niri & e + 2005 - 11 - 06 08:19:53.76 & 11.90 & 2.426453 & 3.398045 & 4.28 & 165.18 & 1.304 & gemini / niri & e + 2005 - 11 - 06 08:31:43.10 & 11.90 & 2.426431 & 3.398051 & 4.28 & 165.19 & 1.254 & gemini / niri & e + 2005 - 12 - 20 10:01:06.82 & 12.35 & 2.598025 & 3.429626 & 10.23 & 141.83 & 1.201 & gemini / niri & e + 2005 - 12 - 20 10:08:25.73 & 12.35 & 2.598075 & 3.429630 & 10.23 & 141.82 & 1.226 &
gemini / niri & e + 2005 - 12 - 21 08:42:06.05 & 12.36 & 2.607454 & 3.430311 & 10.46 & 140.78 & 1.041 & gemini / niri & e + 2005 - 12 - 21 08:47:19.68 & 12.36 & 2.607491 & 3.430314 & 10.46 & 140.78 & 1.048 & gemini / niri & e + 2006 - 01 - 01 10:43:07.10 & 12.56 & 2.731924 & 3.438370 & 12.90 & 128.79 & 1.765 & gemini / niri & e + 2006 - 01 - 06 09:14:36.10 & 12.65 & 2.794520 & 3.441971 & 13.78 & 123.64 & 1.295 & gemini / niri & e + 2006 - 01 - 06 09:20:56.26 & 12.65 & 2.794578 & 3.441974 & 13.78 & 123.64 & 1.323 & gemini / niri & e + 2006 - 12 - 12 16:07:33.89 & 12.73 & 2.883229 & 3.670720 & 10.49 & 137.29 & 1.267 & keck / nirc2 & e + 2010 - 07 - 27 05:23:00.04 & 11.64 & 2.259758 & 3.253753 & 4.03 & 166.98 & 1.027 & vlt / sinfoni & e + 2010 - 08 - 15 08:37:41.66 & 11.94 & 2.331035 & 3.244915 & 8.95 & 150.09 & 1.639 & gemini / niri & e + 2010 - 08 - 25 08:28:46.85 & 12.10 & 2.403504 & 3.240567 & 11.54 & 140.07 & 1.662 & gemini / niri & e + 2010 - 08 - 28 08:22:10.27 & 12.15 & 2.429262 & 3.239301 & 12.25 & 137.09 & 1.671 & gemini / niri & e + 2010 - 08 - 30 05:57:29.00 & 12.18 & 2.446480 & 3.238507 & 12.68 & 135.21 & 1.647 & vlt / ncao & f + 2010 - 09 - 01 08:27:56.74 & 12.22 & 2.466323 & 3.237635 & 13.13 & 133.14 & 1.717 & gemini / niri & e + 2010 - 09 - 02 06:34:03.36 & 12.23 & 2.475252 & 3.237257 & 13.33 & 132.24 & 1.694 & gemini / niri & e + 2011 - 10 - 07 02:37:58.08 & 11.68 & 2.277797 & 3.251690 & 4.40 & 165.53 & 1.357 & vlt / naco & d + 2011 - 11 - 06 02:03:15.84 & 11.99 & 2.376961 & 3.266863 & 8.84 & 149.55 & 1.088 & vlt / naco & d + 2011 - 11 - 08 03:05:36.96 & 12.03 & 2.392516 & 3.267956 & 9.38 & 147.45 & 1.066 & vlt / naco & d + 2011 - 11 - 10 01:04:56.64 & 12.06 & 2.408016 & 3.268988 & 9.88 & 145.48 & 1.149 & vlt / naco & d + 2011 - 11 - 15 00:36:51.84 & 12.15 & 2.452228 & 3.271698 & 11.12 & 140.34 & 1.163 & vlt / naco & d + 2011 - 11 - 16 01:07:06.24 & 12.17 & 2.461963 & 3.272259 & 11.36 & 139.28 & 1.103 & vlt / naco & d + 2011 - 11 - 20 01:06:14.40 & 12.24 & 2.502190 & 3.274473 & 12.27 & 135.18 & 1.086 & vlt / naco & d + 2011 - 12 - 15 05:06:34.56 & 12.65 & 2.816075 & 3.289009 & 16.26 & 110.50 & 1.104 & keck / nirc2 & d + 2011 - 12 - 15 05:29:28.32 & 12.65 & 2.816297 & 3.289018 & 16.27 & 110.48 & 1.087 & keck / nirc2 & d + 2011 - 12 - 15 06:28:30.72 & 12.65 & 2.816873 & 3.289043 & 16.27 & 110.44 & 1.098 & keck / nirc2 & d + 2011 - 12 - 16 04:30:17.28 & 12.66 & 2.829790 & 3.289592 & 16.36 & 109.59 & 1.147 & keck / nirc2 & d + 2011 - 12 - 16 06:01:26.40 & 12.67 & 2.830682 & 3.289630 & 16.36 & 109.53 & 1.084 &
keck / nirc2 & d + 2011 - 12 - 17 06:16:33.60 & 12.68 & 2.844990 & 3.290237 & 16.45 & 108.60 & 1.093 & keck / nirc2 & d + 2011 - 12 - 18 04:43:58.08 & 12.69 & 2.858315 & 3.290799 & 16.53 & 107.74 & 1.114 & keck / nirc2 & d + 2011 - 12 - 18 04:53:02.40 & 12.69 & 2.858404 & 3.290803 & 16.53 & 107.73 & 1.104 & keck / nirc2 & d + 2013 - 01 - 06 01:38:14.18 & 12.30 & 2.654901 & 3.581037 & 6.01 & 157.56 & 1.438 & occ .
data & g + 2001 - 02 - 23t13:40:12.58 & 0.3448 & 0.0290 & 0.3347 & 0.0341 & 0.0101 & -0.0051 & 40.2 & b + 2004 - 08 - 10t07:19:16.00 & 0.3952 & 0.0136 & 0.4046 & 0.0032 & -0.0094 & 0.0104 & 13.3 & c + 2004 - 09 - 01t05:57:01.00 & 0.2074 & -0.0881 & 0.2379 & -0.0936 & -0.0306 & 0.0054 & 13.3 & c + 2004 - 09 - 01t06:05:55.00 & 0.2216 & -0.0904 & 0.2278 & -0.0948 & -0.0063 & 0.0044 & 13.3 & c + 2004 - 09 - 03t07:11:06.00 & -0.2199 & 0.1030 & -0.2082 & 0.0981 & -0.0117 & 0.0049 & 13.3 & c + 2004 - 09 - 05t04:11:28.00 & 0.4052 & -0.0507 & 0.4007 & -0.0489 & 0.0045 & -0.0018 & 13.3 & c + 2004 - 09 - 07t02:38:22.00 & -0.4350 & 0.0009 & -0.4182 & 0.0029 & -0.0168 & -0.0020 & 13.3 & c + 2004 - 09 - 08t06:56:51.00 & -0.2461 & -0.0620 & -0.2451 & -0.0760 & -0.0010 & 0.0141 & 13.3 & c + 2004 - 09 - 13t03:27:24.00 & 0.3693 & 0.0385 & 0.3786 & 0.0298 & -0.0093 & 0.0087 & 13.3 & c + 2004 - 09 - 13t05:32:11.00 & 0.4086 & -0.0136 & 0.4187 & -0.0135 & -0.0101 & -0.0001 & 13.3 & c + 2004 - 09 - 14t03:51:51.00 & -0.1897 & 0.1251 & -0.1793 & 0.1050 & -0.0105 & 0.0202 & 13.3 & c + 2004 - 09 - 15t05:10:21.96 & -0.3956 & 0.0002 & -0.4028 & -0.0109 & 0.0073 & 0.0110 & 13.3 & c + 2004 - 09 - 15t05:18:22.00 & -0.3944 & -0.0012 & -0.4055 & -0.0081 & 0.0112 & 0.0069 & 13.3 & c + 2010 - 08 - 15t08:37:52.00 & 0.2289 & 0.1330 & 0.2249 & 0.1351 & 0.0040 & -0.0021 & 13.3 & e + 2010 - 08 - 28t08:23:47.00 & -0.3366 & -0.0499 & -0.3417 & -0.0461 & 0.0052 & -0.0038 & 13.3 & e + 2010 - 09 - 01t08:29:10.00 & -0.2675 & -0.1232 & -0.2695 & -0.0989 & 0.0020 & -0.0243 & 13.3 & e + 2010 - 09 - 02t06:35:15.00 & 0.3820 & -0.1122 & 0.3555 & -0.1057 & 0.0265 & -0.0066 & 13.3 & e + 2011 - 10 - 07t02:37:58.08 & -0.3910 & -0.0340 & -0.4017 & -0.0277 & 0.0107 & -0.0063 & 13.3 & d + 2011 - 11 - 08t03:05:36.96 & 0.2180 & 0.0570 & 0.2218 & 0.0283 & -0.0038 & 0.0287 & 13.3 & d + 2011 - 11 - 10t01:04:56.64 & -0.3420 & 0.0020 & -0.3564 & -0.0233 & 0.0144 & 0.0253 & 13.3 & d + 2011 - 11 - 16t01:07:06.24 & 0.3810 & 0.0180 & 0.3827 & 0.0126 & -0.0017 & 0.0054 & 13.3 & d + 2011 - 12 - 15t05:06:34.56 & -0.3410 & -0.0110 & -0.3346 & -0.0097 & -0.0064 & -0.0013 & 9.9 & d + 2011 - 12 - 15t05:29:28.32 & -0.3420 & -0.0150 & -0.3347 & -0.0121 & -0.0073 & -0.0029 & 9.9 & d + 2011 - 12 - 15t06:28:30.72 & -0.3250 & -0.0120 & -0.3266 & -0.0179 & 0.0016 & 0.0059 & 9.9 & d + 2011 - 12 - 17t06:16:33.60 & 0.3480 & 0.0350 & 0.3314 & 0.0116 & 0.0166 & 0.0234 & 9.9 & d + 2001 - 02 - 19t08:41:16.998 & 0.3401 & 0.2101 & 0.3589 & 0.2108 & -0.0189 & -0.0007 & 9.9 & a + 2001 - 02 - 19t10:04:18.001 & 0.2806 & 0.2190 & 0.3017 & 0.2178 & -0.0211 & 0.0012 & 9.9 & a + 2001 - 02 - 19t10:13:03.002 & 0.2815 & 0.2421 & 0.2954 & 0.2185 & -0.0139 & 0.0236 & 9.9 & a + 2001 - 02 - 20t09:40:55.997 & -0.6242 & 0.0306 & -0.6273 & 0.0337 & 0.0031 & -0.0031 & 9.9 & a + 2001 - 02 - 20t09:48:26.003 & -0.6161 & 0.0394 & -0.6294 & 0.0317 & 0.0132 & 0.0077 & 9.9 & a + 2001 - 02 - 23t13:40:11.997 & -0.2532 & 0.1957 & -0.2688 & 0.1861 & 0.0156 & 0.0096 & 40.2 & b + 2004 - 07 - 25t10:22:42.003 & -0.7657 & -0.0070 & -0.7530 & -0.0084 & -0.0128 & 0.0015 & 13.3 & c + 2004 - 08 - 10t07:19:16.000 & 0.3738 & 0.1392 & 0.3877 & 0.1408 & -0.0139 & -0.0016 & 13.3 & c + 2004 - 08 - 29t07:20:19.003 & 0.7728 & -0.1082 & 0.7916 & -0.0946 & -0.0188 & -0.0136 & 13.3 & c + 2004 - 08 - 29t07:28:17.996 & 0.7940 & -0.1065 & 0.7890 & -0.0963 & 0.0050 & -0.0102 & 13.3 & c + 2004 - 08 - 29t08:48:08.003 & 0.7434 & -0.1190 & 0.7584 & -0.1132 & -0.0150 & -0.0058 & 13.3 & c + 2004 - 09 - 01t05:57:00.996 & 0.5454 & 0.1366 & 0.5514 & 0.1363 & -0.0060 & 0.0003 & 13.3 & c + 2004 - 09 - 01t06:05:55.000 & 0.5545 & 0.1330 & 0.5581 & 0.1346 & -0.0036 & -0.0016 & 13.3 & c + 2004 - 09 - 01t08:27:37.995 & 0.6583 & 0.1065 & 0.6552 & 0.1057 & 0.0031 & 0.0007 & 13.3 & c + 2004 - 09 - 03t07:11:05.999 & -0.7498 & -0.0610 & -0.7497 & -0.0599 & -0.0001 & -0.0011 & 13.3 & c + 2004 - 09 - 04t08:43:35.999 & -0.1603 & 0.1950 & -0.1423 & 0.2135 & -0.0180 & -0.0185 & 13.3 & c + 2004 - 09 - 05t04:11:27.997 & 0.7851 & 0.0490 & 0.7862 & 0.0433 & -0.0012 & 0.0057 & 13.3 & c + 2004 - 09 - 05t08:17:20.002 & 0.8362 & -0.0192 & 0.8363 & -0.0191 & -0.0001 & -0.0002 & 13.3 & c + 2004 - 09 - 06t03:52:03.999 & 0.1661 & -0.2058 & 0.1858 & -0.2126 & -0.0197 & 0.0068 & 13.3 & c + 2004 - 09 - 07t02:38:22.004 & -0.8307 & 0.0053 & -0.8216 & 0.0046 & -0.0091 & 0.0007 & 13.3 & c + 2004 - 09 - 07t09:26:55.999 & -0.7914 & 0.1163 & -0.7785 & 0.1060 & -0.0129 & 0.0104 & 13.3 & c + 2004 - 09 - 08t06:56:51.002 & 0.2542 & 0.1941 & 0.2694 & 0.1926 & -0.0152 & 0.0015 & 13.3 & c + 2004 - 09 - 11t04:53:29.996 & -0.6726 & 0.1650 & -0.6542 & 0.1600 & -0.0184 & 0.0050 & 13.3 & c + 2004 - 09 - 13t03:27:24.001 & 0.5571 & -0.1980 & 0.5477 & -0.1852 & 0.0094 & -0.0128 & 13.3 & c + 2004 - 09 - 13t05:32:11.002 & 0.4478 & -0.1852 & 0.4482 & -0.2004 & -0.0004 & 0.0153 & 13.3 & c + 2004 - 09 - 14t03:51:50.996 & -0.7220 & -0.0725 & -0.7136 & -0.0758 & -0.0084 & 0.0033 & 13.3 & c + 2004 - 09 - 14t06:38:29.999 & -0.7710 & -0.0361 & -0.7811 & -0.0330 & 0.0101 & -0.0031 & 13.3 & c + 2004 - 09 - 14t06:46:34.003 & -0.7829 & -0.0314 & -0.7836 & -0.0309 & 0.0007 & -0.0005 & 13.3 & c + 2004 - 09 - 15t05:18:22.003 & -0.2080 & 0.2215 & -0.2048 & 0.2204 & -0.0032 & 0.0011 & 13.3 & c + 2004 - 10 - 19t00:29:36.003 & 0.7423 & -0.0400 & 0.7424 & -0.0392 & -0.0001 & -0.0008 & 13.3 & c + 2004 - 10 - 20t00:10:57.996 & -0.0891 & -0.1748 & -0.0993 & -0.1846 & 0.0101 & 0.0098 & 13.3 & c + 2004 - 10 - 25t06:29:31.001 & -0.1298 & 0.1851 & -0.1333 & 0.1891 & 0.0034 & -0.0040 & 9.9 & c + 2004 - 11 - 02t07:35:11.999 & 0.6486 & 0.0460 & 0.6243 & 0.0427 & 0.0244 & 0.0033 & 22.0 & c + 2005 - 11 - 01t13:20:57.001 & 0.0015 & -0.2670 & -0.0046 & -0.2581 & 0.0061 & -0.0089 & 22.0 & e + 2005 - 11 - 06t08:20:06.002 & 0.6840 & 0.1901 & 0.6911 & 0.1814 & -0.0071 & 0.0088 & 22.0 & e + 2005 - 11 - 06t08:32:42.996 & 0.6914 & 0.1739 & 0.6862 & 0.1844 & 0.0051 & -0.0105 & 22.0 & e + 2005 - 12 - 20t10:01:09.001 & 0.2384 & 0.2573 & 0.2450 & 0.2594 & -0.0066 & -0.0021 & 22.0 & e + 2005 - 12 - 20t10:10:25.996 & 0.2434 & 0.2434 & 0.2375 & 0.2596 & 0.0059 & -0.0162 & 22.0 & e + 2005 - 12 - 21t08:42:09.002 & -0.7005 & 0.0065 & -0.6844 & 0.0035 & -0.0161 & 0.0030 & 22.0 & e + 2005 - 12 - 21t08:48:50.002 & -0.6704 & 0.0129 & -0.6860 & 0.0014 & 0.0156 & 0.0115 & 22.0 & e + 2006 - 01 - 01t10:44:37.003 & -0.6687 & -0.0559 & -0.6808 & -0.0677 & 0.0121 & 0.0118 & 22.0 & e + 2006 - 01 - 06t09:14:37.996 & 0.4125 & -0.1554 & 0.4061 & -0.1544 & 0.0063 & -0.0009 & 22.0 & e + 2006 - 01 - 06t09:22:55.997 & 0.4112 & -0.1418 & 0.4113 & -0.1525 & -0.0000 & 0.0106 & 22.0 & e + 2006 - 12 - 12t16:09:34.001 & 0.5659 & 0.1796 & 0.5353 & 0.1994 & 0.0306 & -0.0199 & 9.9 & e + 2010 - 07 - 27t05:23:00.040 & 0.3900 & 0.2950 & 0.3719 & 0.3024 & 0.0181 & -0.0074 & 12.5 & e + 2010 - 08 - 15t08:37:51.997 & 0.6981 & -0.2980 & 0.6827 & -0.2803 & 0.0155 & -0.0177 & 22.0 & e + 2010 - 08 - 25t08:30:23.002 & 0.4274 & 0.2690 & 0.4061 & 0.2778 & 0.0213 & -0.0089 & 22.0 & e + 2010 - 08 - 28t08:23:46.996 & -0.4167 & 0.3680 & -0.3989 & 0.3656 & -0.0178 & 0.0024 & 22.0 & e + 2010 - 08 - 30t05:57:28.998 & 0.3225 & -0.3714 & 0.3252 & -0.3766 & -0.0027 & 0.0052 & 13.3 & f + 2010 - 09 - 01t08:29:10.003 & 0.0744 & 0.3535 & 0.0584 & 0.3638 & 0.0161 & -0.0102 & 22.0 & e + 2010 - 09 - 02t06:35:15.002 & 0.7748 & -0.1153 & 0.7493 & -0.1171 & 0.0256 & 0.0018 & 22.0 & e + 2011 - 10 - 07t02:37:58.002 & -0.6980 & -0.0720 & -0.7047 & -0.0615 & 0.0067 & -0.0105 & 13.3 & d + 2011 - 11 - 06t02:03:15.001 & 0.3680 & -0.0190 & 0.3771 & -0.0250 & -0.0091 & 0.0060 & 13.3 & d + 2011 - 11 - 08t03:05:36.000 & -0.6060 & 0.0150 & -0.6088 & 0.0054 & 0.0028 & 0.0096 & 13.3 & d + 2011 - 11 - 10t01:04:56.000 & 0.6430 & 0.0120 & 0.6741 & 0.0015 & -0.0311 & 0.0105 & 13.3 & d + 2011 - 11 - 15t00:36:51.001 & -0.1770 & 0.0460 & -0.1824 & 0.0373 & 0.0054 & 0.0087 & 13.3 & d + 2011 - 11 - 16t01:07:05.998 & -0.6700 & -0.0450 & -0.6832 & -0.0423 & 0.0132 & -0.0027 & 13.3 & d + 2011 - 11 - 20t01:06:14.002 & -0.3550 & -0.0350 & -0.3609 & -0.0527 & 0.0059 & 0.0177 & 13.3 & d + 2011 - 12 - 15t05:06:33.998 & -0.5680 & -0.0570 & -0.5736 & -0.0483 & 0.0056 & -0.0087 & 9.9 & d + 2011 - 12 - 15t05:29:28.000 & -0.5680 & -0.0630 & -0.5644 & -0.0495 & -0.0036 & -0.0135 & 9.9 & d + 2011 - 12 - 15t06:28:30.002 & -0.5390 & -0.0460 & -0.5387 & -0.0523 & -0.0003 & 0.0063 & 9.9 & d + 2011 - 12 - 16t04:30:17.003 & 0.3910 & -0.0370 & 0.3829 & -0.0392 & 0.0081 & 0.0022 & 9.9 & d + 2011 - 12 - 16t06:01:26.002 & 0.4460 & -0.0160 & 0.4385 & -0.0334 & 0.0075 & 0.0174 & 9.9 & d + 2011 - 12 - 17t06:16:33.003 & 0.4150 & 0.0820 & 0.4080 & 0.0612 & 0.0070 & 0.0208 & 9.9 & d + 2011 - 12 - 18t04:43:58.002 & -0.5280 & 0.0260 & -0.5306 & 0.0200 & 0.0026 & 0.0060 & 9.9 & d + 2011 - 12 - 18t04:53:02.002 & -0.5210 & 0.0190 & -0.5346 & 0.0193 & 0.0136 & -0.0003 & 9.9 & d + 2013 - 01 - 06t01:38:14.000 & 0.5930 & -0.1130 & 0.5898 & -0.1041 & 0.0032 & -0.0089 & 4.5 & g + s & a. carbognani & 17.70 & 01:37:33.30 @xmath600.28 & 01:37:51.00 @xmath600.28 & + s & s. bolzoni & 18.20 & 01:37:29.15 @xmath600.30 & 01:37:47.35 @xmath600.20 & + s & s. sposetti , a. manna & 25.34 & 01:37:26.21 @xmath600.03 & 01:37:51.55 @xmath600.04 & + s & v. fristot & 21.50 & 01:37:30.70 @xmath600.20 & 01:37:52.20 @xmath600.20 & + s & a. figer & 26.20 & 01:37:22.80 @xmath600.60 & 01:37:49.00 @xmath600.80 & + s & m. bretton & 18.10 & 01:37:34.50 @xmath600.0 ^b^ & 01:37:52.60 @xmath600.0 ^b^ & -3.8 + s & j. lecacheux & 4.06 & 01:37:37.16 @xmath600.08 & 01:37:41.22 @xmath600.10 & + s & p. tanga & 19.19 & 00:00:00.00 @xmath600.02 ^c^ & 00:00:19.19 @xmath600.02 ^c^ & + s & m. devogele et al . &
12.77 & 01:37:26.73 @xmath600.08 & 01:37:39.50 @xmath600.08 & + s & e. frappa , a. klotz & 12.90 & 01:37:26.60 @xmath600.10 & 01:37:39.50 @xmath600.10 & + s & l. brunetto et al . & 7.12 & 01:37:29.53 @xmath600.02 & 01:37:36.65 @xmath600.02 & + s & v. metallinos & 24.60 & 01:36:20.73 @xmath600.05 & 01:36:45.33 @xmath600.05 & + & & & & & + r & p. tanga & 1.96 & 00:00:97.10 @xmath600.03 ^c^ & 00:00:99.06 @xmath600.03 ^c^ & + r & p. dubreuil & 1.70 & 01:38:57.60 @xmath602.00 & 01:38:59.30 @xmath602.00 & 0.5 + r & m. devogele et al . & 0.56 & 01:39:00.85 @xmath600.08 & 01:39:01.41 @xmath600.08 & + r & e. frappa , a. klotz & 0.48 & 01:39:00.73 @xmath600.08 & 01:39:01.21 @xmath600.08 & + s. sposetti ( ch ) & m400 , vid & e 09 01 26.5 & n 46 13 53.2 & 260 & ws + c. gualdoni ( it ) & m250 , vid & e 09 06 01.0 & n 45 48 18.0 & 255 & ws + a. carbognani ( it ) & m810 , ccd , ntp & e 07 28 42.0 & n 45 47 23.5 & 1678 & ws + s. bolzoni ( it ) & m300 , vis , rad+ & e 08 09 59.8 & n 45 21 57.0 & 195 & ws + u. quadri et al .
( it ) & m250 , ccd & e 10 07 49.5 & n 45 19 32.4 & 63 & ws + s. sposetti , a. manna ( it ) & m200 , vid , gps++ & e 07 18 01.3 & n 45 07 34.1 & 585 & ws + v. fristot ( fr ) & l102 , ccd , ntp & e 05 16 35.4 & n 44 41 35.5 & 390 & ws + a. figer ( fr ) & l68 , ccd , phone & e 06 40 54.0 & n 44 34 24.0 & 1845 & ws + m. bretton ( fr ) & m820 , vis , rad & e 05 30 54.4 & n 44 24 29.9 & 810 & ws + j .- l .
penninckx ( fr ) & m400 , vis & e 04 22 45.8 & n 44 07 51.5 & 245 & + c. peguet ( fr ) & m350 , vis & e 05 00 09.0 & n 44 07 31.8 & 94 & ws + l. bernasconi ( fr ) & m500 , ccd & e 05 11 11.2 &
n 44 01 17.5 & 330 & ws + j. lecacheux ( fr ) & l130 , vid , gps++ & e 06 01 15.3 & n 43 58 48.6 & 720 & ws + e. frappa et al . ( fr ) & l61 , vid & e 05 58 33.4 & n 43 56 07.9 & 460 & ws + l. arnold ( fr ) & m205 , ccd & e 05 42 48.4 &
n 43 55 00.7 & 551 & ws + e. frappa ( fr ) & m203 , vid & e 06 00 23.0 & n 43 51 51.9 & 628 & ws + g. brabant ( fr ) & m200 , vid & e 04 54 47.5 &
n 43 47 48.8 & 53 & ws + p. tanga ( fr ) & m356 , ccd & e 07 15 47.2 & n 43 47 22.2 & 385 & ws + p. dubreuil ( fr ) & m203 , ccd & e 07 14 30.6 & n 43 46 58.5 & 480 & ws + d. verilhac ( fr ) & m210 , vid & e 04 53 35.1 & n 43 46 01.0 & 115 & ws
+ m. devogele et al .
( fr ) & m1000 , vid , gps++ & e 06 55 21.8 & n 43 45 13.5 & 1280 & ws + e. frappa , a. klotz ( fr ) & m250 , ccd , gps++ & e 06 55 25.1 & n 43 45 07.3 & 1270 & ws + o. lecacheux ( fr ) & l61 , vid , gps++ & e 05 30 00.0 & n 43 41 31.6 & 214 & ws + r. poncy ( fr ) & m400 , ccd & e 03 56 24.0 & n 43 38 50.0 & 54 & ws + l. brunetto et al . ( fr ) & m406 , vid , gps++ & e 07 04 18.4 & n 43 36 15.7 & 130 & ws + d. albanese ( fr ) & m280 , ccd & e 06 39 04.9 & n 43 29 45.9 & 43 & ws + f. colas ( fr ) & m1050 , ccd & e 00 08 32.5 & n 42 56 10.9 & 2871 & ws + j. lopez ( es ) & m200 , vid & e 02 59 50.5 &
n 42 15 00.2 & 19 & w + c. perello , a. selva ( es ) & m500 , vid & e 02 05 24.6 & n 41 33 00.2 & 224 & ws + r. casas ( es ) & m200 , vid & e 02 07 14.3 & n 41 32 22.1 & 165 & ws + h. pallares ( es ) & m280 , vis & e 04 14 28.1 & n 39 49 12.7 & 19 & w + v. metallinos ( gr ) & l130 , vid , gps++ & e 19 52 23.3 & n 39 38 07.4 & 2 & ws + spin axis ecliptic j2000 longitude & 70 & @xmath81 + spin axis ecliptic j2000 latitude & 69 & @xmath82 + sidereal rotation period & 5.183640 h & @xmath83 h + volume - equivalent diameter @xmath3 & 273 km & @xmath84 km + quadrupole term @xmath5 & 0.024 & 0.015 0.040 + volume & @xmath85 km@xmath86 & @xmath87 km@xmath86 + surface area & @xmath88 km@xmath89 & @xmath90 km@xmath89 + + @xmath91 & 193 km & 180 210 km + @xmath92 & 129 km & 127 160 km + @xmath93 & 124 km & 115 150 km + + @xmath94 & 1.46 & 1.30 1.75 + @xmath95 & 1.07 & 1.02 1.20 + @xmath96 ( day ) & 1.356654 & @xmath60 @xmath97 & 3.641191 & @xmath60 @xmath98 + @xmath91 ( km ) & 684.4 & @xmath60 109.7 & 1351.7 & @xmath60 151.0 + @xmath99 & 0.0 & @xmath60 0.05 & 0.007 & @xmath60 0.047 + @xmath100 ( deg ) & 8.6 & @xmath60 13.5 & 8.3 & @xmath60 7.7 + @xmath101 ( deg ) & 93.0 & @xmath60 48.0 & 93.6 & @xmath60 20.1 + @xmath102 ( deg ) & 187.9 & @xmath60 52.1 & 109.1 & @xmath60 23.6 + @xmath103 ( jd ) & 2455594.58824 & @xmath60 0.158 & 2455596.41837 & @xmath60 0.202 + + @xmath104 ( @xmath105 kg ) & 1.380 & @xmath60 0.669 & 1.476 & @xmath60 0.497 + @xmath106 ( g.@xmath58 ) & 1.29 & @xmath60 0.59 & 1.38 & @xmath60 0.44 + @xmath107 ( deg ) & 70 , + 65 & @xmath60 30 , 9 & 70 , + 65 & @xmath60 17 , 4 + 2016 - 01 - 30 07:22 & 6226 01275 & 12.1 & 1.6 & 8.0 & brazil + 2016 - 06 - 22 23:54 & 6815 03609 & 11.6 & 0.6 & 20.5 & chile , argentina , south africa , madagascar + 2016 - 08 - 16 23:36 & 6817 01360 & 12.6 & 1.1 & 44.6 & chile , paraguay , brazil ( south ) + 2016 - 11 - 03 23:17 & 6854 00301 & 10.6 & 1.4 & 7.2 & chile , argentina , brazil ( south ) + 2019 - 10 - 15 08:11 & 1931 01512 & 9.9 & 2.0 & 14.1 & chile , argentina , brazil ( south ) , south africa + 2019 - 10 - 20 06:08 & 1932 00479 & 12.0 & 1.9 & 15.8 & spain , france ( south ) , italy , greece + 2019 - 10 - 29 23:45 & 1932 00469 & 10.0 & 1.9 & 21.0 & spain , france , italy , germany , poland + 2023 - 07 - 06 08:01 & 7442 01392 & 11.9 & 1.0 & 22.2 & mexico , usa ( east ) , canada ( east ) + ) . the high levels of intensity showing the irregular shape of the primary are shown in the central circle .
north is up , and east is left .
the plate scale is 40.2 mas for the hst / wfpc2 , 9.9 mas for the keck / nirc2 , and 13.3 mas for the vlt / naco observations .
[ fig - ao ] ] and @xmath108 of the sub - earth and sub - solar point , respectively , and by the solar phase angle @xmath109 .
the observed contours of ao images and the corresponding projections of the model ( smooth blue curves ) are plotted in the second panel . in the third panel ,
occultation chords and the model outline are shown .
the dashed lines are negative observations.[fig - colfits ] ] .
the dashed black lines are negative observations ( northern one by v. fristot ) .
the blue ( leftmost ) and red lines show the predicted orbit of romulus projected into the occultation plane , respectively based on the previous and the new orbital models .
the two marks on these lines ( located at @xmath110 km ) enlighten the predicted positions of romulus at the epoch of occultation . [ fig - romulusbestmodel ] ] with _ genoid - anis_. the gray dots are all solutions .
the ( blue ) biggest points show the solutions for which the orbital pole covers a range of @xmath111 around the best - fit spin solution of the primary ( i.e. , rectangles in fig .
[ fig - rmspole ] ) .
the ( green , blue , red ) horizontal dashed lines correspond , respectively , to 1@xmath34 , 2@xmath34 and 3@xmath34 thresholds .
the three vertical lines indicate the @xmath5 values estimated , from left to right , from the 3-d shape model ( see sec .
[ sylvia ] ) , by @xcite and by @xcite .
the best candidate solutions are obtained for @xmath63.[fig - j2 ] ] 67 natexlab#1#1url # 1`#1`urlprefix[2][]#2 [ 2]#2 , , , , , , , , , , , . .
, , , , , , , , , , , , , , . .
, , , , , . . ,
, , , , . , .
, , , , , , , , , , , , , , , . , in : , p. .
, , , , , , , , , , , , , , , . . ,
, , , , , . .
, , , , , , , , , , , , , , , , , , , , , , , . . ,
. , . . , . , , , , , , , , , , ,
. . , . , , , , , , , , , , , , , , ,
. . , . , , , , , , , , , , , , , , , , , , ,
. . , . , , , , , , , , , , , , , , , , , , ,
. . , . , . , . , ,
. . , . , , , , , ,
, , , , , , , , , , , , , , , , , , , , , , , . , in : , ( eds . ) , , pp . . , , , , , , , , , , , , , , , , , , . , in : . , , , , , , . .
, , , . . ,
, , , . . ,
, , , . . ,
, , , a. . , .
, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , b. . , . , , .
, , , , , , , , , , , , , , , , , , , , , , . , in : ( ed . ) ,
, pp . . , , , , , , , , , , , , , ,
. . , . , , ,
. . , . , , , , , , , . , , , , , , , , , ,
. , in : , ( eds . ) , , pp . . , , , , , , , a. . , .
, , , , a. . , . , , , , , , b. . , . , , , , , , , , b. . , . , , , , , , , , , , , , .
, , , , a. . .
, , , , , , , , , , , , , , b. . , .
, , , , , , , . . ,
, , , , , , , , , , , , , . . ,
. , . . , . , , , ,
, , , . . ,
, , , , , , , , , , , , , , , , , , . , in : , ( eds . ) , , pp . . , , . .
, , , , , , , . . ,
, , , , . . ,
, , , , , , , . .
, , , , , . . ,
, , , , , , , , , , , , . . ,
, , , . . ,
, , , . . ,
, , , , , , , . . ,
, , , , , , , , , , . . ,
, , , , . . ,
, , , . . ,
, , , , , . . ,
, , , , , , . . ,
, , , , , , , , . . ,
, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , . , in : . , , , , , , . , in : ( ed
. ) , , pp . . | we present the analysis of high angular resolution observations of the triple asteroid ( 87 ) sylvia collected with three 810 m class telescopes ( keck , vlt , gemini north ) and the hubble space telescope .
the moons mutual orbits were derived individually using a purely keplerian model .
we computed the position of romulus , the outer moon of the system , at the epoch of a recent stellar occultation which was successfully observed at less than 15 km from our predicted position , within the uncertainty of our model .
the occultation data revealed that the moon , with a surface - area equivalent diameter @xmath0 km , is strongly elongated ( axes ratio of @xmath1 ) , significantly more than single asteroids of similar size in the main - belt .
we concluded that its shape is probably affected by the tides from the primary .
a new shape model of the primary was calculated combining adaptive - optics observations with this occultation and 40 archived light - curves recorded since 1978 .
the difference between the @xmath2 derived from the 3-d shape model assuming an homogeneous distribution of mass for the volume equivalent diameter @xmath3= @xmath4 km primary and the null @xmath5 implied by the keplerian orbits suggests a non - homogeneous mass distribution in the asteroid s interior .
asteroids , satellites of asteroids , adaptive optics , photometry , orbit determination , occultations |
t tauri stars are low - mass ( k & m spectral types ) young stars which are in their pre - main sequence phase of evolution ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
they are often associated with cloud complexes such as taurus , orion and ophiuchus @xcite . however
, t tauri stars are also found in isolated regions above the galactic plane and far from any dark clouds @xcite .
tw hydrae was the first such ` isolated t tauri star ' identified , which was later found to be part of an association of about three dozen members , known as tw hydrae association ( twa , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
they were not runaway stars from molecular clouds but formed in situ in the present region @xmath410 myr ago , which is now devoid of molecular gas @xcite . since then several such nearby ,
young associations have been identified such as , the @xmath5 pictoris moving group , the ab doradus moving group , the tucana / horologium association etc . ,
within 100 pc of the sun ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for a review ) .
in this paper we present a detailed analysis of the high galactic latitude ( b = @xmath030@xmath6 ) binary t tauri star system pds 11 ( gsc 04744 - 01367 , iras 04451 - 0539 ) .
@xcite have reported this to be a binary system with components pds ( pds stands for pico dos dias survey ) 11a and pds 11b .
pds 11a is the north - eastern component of the binary ( figure 1 ) .
pds 11a and pds 11b are present in the washington double star ( wds ) catalog , with the component magnitudes being 14.76 and 15.34 , respectively .
the position angle ( pa ) and separation were found to be 216@xmath6 and 8.8@xmath7 , respectively @xcite .
our analysis indicate that pds 11 is a young ( 10@xmath015 myr ) , nearby ( 114@xmath0131 pc ) binary t tauri system , where both the components are possibly accreting . the paper is arranged as follows .
section 2 describes the optical and near - ir observations .
the results from this study are presented in sect .
3 . a detailed discussion of the key results is given in sect . 4 , and finally in sect .
5 we present our conclusions .
optical spectra of both the components of the pds 11 system in the wavelength range 3800@xmath09000 were obtained with the himalayan faint object spectrograph camera ( hfosc ) mounted on the 2-m himalayan chandra telescope ( hct ) .
the wavelength range was covered using grism 7 ( blue region , 3800@xmath05500 ) and grism 8 ( red region , 5500@xmath09000 ) .
the grisms in combination with 1.92@xmath7 wide and 11@xmath8 long slit provide an effective resolving power of @xmath4 900 in blue region and @xmath4 1050 in red region .
a spectrophotometric standard , feige 34 , was observed on the same night and is used for flux calibration .
the observations were carried out on 2016 february 13 .
the fene , fear lamp spectra were taken after each on - source exposure for wavelength calibration . the spectra were reduced in a standard manner after bias subtraction and flat field correction using the standard tasks in image reduction and analysis facility ( iraf ) .
further , the extracted spectra were wavelength calibrated and flux calibrated .
the target was again observed on march 15 to confirm the spectral features observed .
the log of observations is given in table 1 .
we also obtained near - ir spectra of pds 11 with tifr near - infrared spectrometer and imager ( tirspec ; * ? ? ? * ) mounted on 2-m hct .
the spectra were taken in @xmath9 ( [email protected] @xmath10 m ) , @xmath11 ( [email protected] @xmath10 m ) , @xmath12 ( [email protected] @xmath10 m ) and @xmath13 ( [email protected] @xmath10 m ) bands , in combination with grism and l3 slit ( 1.97@xmath7 wide and 300@xmath7 long ) .
the effective resolving power is around 1200 .
the program stars pds 11a and pds 11b are co - aligned in the slit and care is taken while reducing the spectra to extract them separately . a telluric standard hip34768 ( a1v spectral type ) at nearby airmass was also observed .
the observations were carried out in dithered mode .
the log of infrared spectroscopic observations is given in table 1 .
argon spectra were obtained after the object spectra for wavelength calibration .
the wavelength calibrated object spectrum is divided with the the telluric spectrum whose hydrogen absorption lines were removed .
the resultant spectrum is multiplied with a blackbody spectrum of 9230 k , corresponding to a1v spectral type of the telluric standard @xcite .
the final spectrum is normalized with respect to the band center in @xmath9 , @xmath11 , @xmath12 and @xmath13 bands .
llcccccc pds 11a & 2016 feb .
13 & 1800 & 1800 & . & . & . & .
+ & 2016 feb .
& 1000 & 1000 & 1000 & 1000 + & 2016 mar .
15 & 1800 & 1800 & . & . & . & .
+ pds 11b & 2016 feb .
13 & 1800 & 1800 & . & . & . & .
+ & 2016 feb .
& 1000 & 1000 & 1000 & 900 + & 2016 mar .
15 & 900 & 900 & . & . & . & .
+ feige 34 & 2016 feb .
13 & 600 & 600 & . & . & . & .
+ hip 34768 & 2016 feb .
& 400 & 400 & 240 & 160 + we imaged the 10@xmath8@xmath1410@xmath8 region centered on pds 11 in @xmath15 passbands @xcite on 2016 march 19 using hfosc .
the data reduction was carried out using various packages available in iraf .
aperture photometry was performed on the program star and nearby field stars .
the @xmath16 , @xmath17 , @xmath18 magnitudes of the nearby field stars were obtained using available sdss photometry , which were converted to bessell system using the transformation relations given in lupton ( 2005 ) .
the magnitude of the program stars were calibrated differentially with respect to the nearby field stars .
the spectra of both pds 11a ( figure 2 ) and pds 11b ( figure 3 ) look very similar due to the presence of balmer emission lines , from h@xmath1 all the way up to h8 ( 8@xmath02 ) , caii h & k emission lines and tio absorption bands , albeit with different line strengths .
@xcite reported an h@xmath1 equivalent width ( ew ) of @xmath020 for pds 11a and @xmath042 for pds 11b . the equivalent width measured from our spectra
are @xmath023 and @xmath027 , which is quite different from the earlier measurements , particularly for pds 11b .
the ew and full width at half maximum ( fwhm ) of the prominent spectral lines are given in table 2 .
we found [ oi ] @xmath26300 in emission in pds 11b ( figure 4 ) .
this line is often seen in the collimated jets driven by accreting t tauri stars ( e.g. * ? ? ?
* ; * ? ? ?
@xcite also noticed [ oi ] @xmath26300 in the spectrum of pds 11b .
this emission line is not seen in the spectrum of pds 11a , both in our observations and that of @xcite . @xcite
have suggested that [ oi ] @xmath26300 is formed by the photodissociation of oh molecules in the disk by far - ultraviolet stellar photons .
we find evidence for the presence of hei @xmath24471 and @xmath25876 in emission in the spectra of pds 11b ( figure 3 ) .
hei @xmath24471 is red shifted by 3 , as seen from the spectra observed on 2016 february 13 .
this feature was absent during the observation on march 15 .
hei @xmath25876 is affected by molecular absorption bands . however ,
when compared with the spectrum of pds 11a , where hei @xmath25876 is equally affected by molecular band absorption , an emission component is seen at @xmath25876 . it may be noted that @xcite identified hei @xmath25876 emission in pds 11b .
it is possible that hei emission lines are formed from photoionization and subsequent recombination in the accretion shock region , close to the stellar surface @xcite .
hence , the presence of [ oi ] @xmath26300 and hei emission lines in the spectrum of pds 11b suggest that it belongs to the class of accreting t tauri stars .
the flux calibrated @xmath9 , @xmath11 , @xmath12 , @xmath13 spectra of pds 11a and pds 11b , normalized to band center values are shown in figure 5 .
since the signal to noise is low , the spectra is smoothed to 5 points for display purpose .
pa@xmath5 is the only prominent spectral line found in pds 11a , which is present in emission with an ew of 3 .
no prominent emission or absorption features are present in pds 11b .
we estimated the spectral type of pds 11a and pds 11b using tio bands , the dominant molecular absorption band in m - type stars .
the spectral type has been estimated from the tio5 spectral index , which is defined as the ratio of mean flux value in 7126@xmath07135 region to that at 7042@xmath07046 wavelength region @xcite .
@xcite derived a relation between the spectral type ( spt ) and tio5 index , spt = @xmath010.775@xmath14tio5 + 8.2 .
they suggested using this relation as a reliable spectral type estimator in the range k7 to m6.5 dwarfs .
we have measured the flux values from the flux calibrated spectra obtained on 2016 february 13 and found tio5 to be [email protected] for pds 11a and [email protected] for pds 11b .
the spectral type is estimated from the above mentioned relation whereby pds 11a is found to be [email protected] and pds 11b to be [email protected] . since fractional subtypes are not used for analysis , we will be approximating the spectral type of both stars as m2 , but suggest that pds 11a can be earlier than pds 11b by a fractional subtype .
further , we compared the absorption strength of tio @xmath26180 absorption feature with the m2v spectral template from pickles library ( see figure 4 ) .
a close match between the tio band strength of the object and the template spectrum support the m2 classification of pds 11a and pds 11b .
the early m - type classification is further supported by the absence of tio @xmath28465 band in our spectra .
this band starts appearing in the spectra from m3 onwards and is used as a criteria to classify objects with spectral type m3@xmath0l3 @xcite .
llcccc caii k & 2016 feb .
13 & [email protected] & [email protected] & -48@xmath194 & [email protected] + & 2016 mar .
15 & -28@xmath193 & 9.5@xmath191 & -38@xmath196 & 7.5@xmath192 + caii h + h@xmath20 & 2016 feb .
13 & -22@xmath191 & [email protected] & -23@xmath193 & [email protected] + & 2016 mar .
15 & -17@xmath192 & [email protected] & -24@xmath194 & 8.0@xmath191 + h@xmath21 & 2016 feb . 13 & [email protected] & [email protected] & [email protected] & [email protected] + & 2016 mar .
15 & [email protected] & [email protected] & -10@xmath193 & 8.0@xmath192 + h@xmath22 & 2016 feb .
13 & [email protected] & [email protected] & [email protected] & [email protected] + & 2016 mar .
15 & [email protected] & 9.0@xmath191 & -11@xmath192 & 8.5@xmath191 + 4474 ( hei ? ) & 2016 feb .
13 & . & . &
[email protected] & [email protected] + & 2016 mar .
15 & . & . & . & .
+ h@xmath5 & 2016 feb .
13 & [email protected] & [email protected] & [email protected] & [email protected] + & 2016 mar .
15 & [email protected] & [email protected] & -10@xmath191 & [email protected] + hei 5876 & 2016 feb . 13 & . & . &
[email protected] & [email protected] + & 2016 mar .
15 & . & . &
[email protected] & [email protected] + nai ( 5890 + 5896 ) & 2016 feb .
13 & [email protected] & [email protected] & [email protected] & [email protected] + & 2016 mar .
15 & [email protected] & [email protected] & [email protected] & [email protected] + [ oi]6300 & 2016 feb .
13 & . & . &
[email protected] & [email protected] + & 2016 mar .
15 & . & . &
[email protected] & [email protected] + h@xmath1 & 2016 feb .
13 & [email protected] & [email protected] & [email protected] & [email protected] + & 2016 mar .
15 & [email protected] & [email protected] & [email protected] & [email protected] + lii 6708 & 2016 feb .
13 & [email protected] & [email protected] & . & .
+ & 2016 mar .
15 & [email protected] & [email protected] & . &
+ nai 8183 & 2016 feb .
13 & [email protected] & [email protected] & [email protected] & [email protected] + & 2016 mar .
15 & [email protected] & [email protected] & [email protected] & [email protected] + nai 8195 & 2016 feb .
13 & [email protected] & [email protected] & [email protected] & [email protected] + & 2016 mar .
15 & [email protected] & [email protected] & [email protected] & [email protected] + historically , the strength of the h@xmath1 line has been used to distinguish between accreting classical t tauri stars ( ctts ) from weak - lined t tauri stars ( wtts ) , where the h@xmath1 emission is due to chromospheric activity .
an h@xmath1 equivalent width ew(h@xmath1 ) @xmath4 10 was set as the discrimination boundary .
however , because of the ` contrast effect ' of the photosphere , no unique ew(h@xmath1 ) value distinguishes all cttss from wttss , and several authors have proposed ew(h@xmath1 ) dividing line as function of spectral type @xcite .
@xcite prescribed empirically determined maximum ew(h@xmath1 ) values observed for non - accreting t tauri stars for different spectral type ranges .
@xcite proposed ew(h@xmath1 ) values as a function of spectral type derived from the observed saturation limit for the chromospheric activity at log(l@xmath23/l@xmath24 ) = @xmath03.3 .
figure 6 shows both @xcite and @xcite criteria to distinguish between cttss from wttss .
also shown in the figure are the ew(h@xmath1 ) values of pds 11a and pds 11b , which are well above the lines depicting @xcite and @xcite criteria .
thus , both pds 11a and pds 11b are accreting and are cttss .
we computed the accretion rates for pds 11a from the h@xmath1 and h@xmath5 line luminosities using the empirical relations given by @xcite , @xcite , and @xcite . for the distance range listed in table 3 ( 114@xmath0131 pc )
corresponding to the age range of 10@xmath015 myr , the accretion rates obtained are in the range of 4.2@xmath1410@xmath25 @xmath0 5.0@xmath1410@xmath26 m@xmath3 yr@xmath27 with a median value of @xmath4 1.0@xmath1410@xmath26 m@xmath3 yr@xmath27 .
the accretion rates we obtain for pds 11a is significantly lower than that found for @xmath28 3 myr old cttss of spectral type m2 ( e.g. , * ? ? ?
* ; * ? ? ?
they are , however , quite similar to the accretion rates found for m2 members of the 10@xmath015 myr old moving groups ( e.g. , * ? ? ?
the hi line luminosity and the @xmath29 estimated from them for the pds 11a are significantly higher than that expected from chromospheric activity . following @xcite , the noise introduced in the estimated @xmath29 due to chromospheric contamination ( see eqn . 2 in * ? ? ?
* ) is @xmath28 1.8@xmath1410@xmath30 @xmath31 , while the accretion luminosity of pds 11a is @xmath4 1.4@xmath1410@xmath32 @xmath31 , which is @xmath4 8 times higher , indicating that pds 11a is accreting material from the disk . from the accretion rates of pds 11a
, we further estimated the expected line flux and equivalent widths for pa@xmath5 and br@xmath22 lines using the empirical relations from @xcite , @xcite and @xcite . for pds 11a , the expected pa@xmath5 ew is @xmath42 ( line flux @xmath42.1@xmath1410@xmath33 w @xmath34 ) and the expected br@xmath22 ew is @xmath40.3 ( line flux @xmath43.1@xmath1410@xmath35 w @xmath34 ) . from the observed spectra discussed in sect .
3.1 we found that pa@xmath5 ew is around 3 whereas no clear emission is present in br@xmath22 , which agrees with these estimates .
thus the strength of the observed hi lines in the optical and near - ir spectra of pds 11a is consistent with the star accreting at a rate of @xmath410@xmath26 m@xmath3 yr@xmath27 .
the lii @xmath26708 absorption feature is an indicator of youth and is often used as one of the criteria to classify the source as a t tauri star @xcite .
we found evidence for the presence of lii @xmath26708 line in the spectrum of pds 11a , with an ew of 0.43 ( mean value of both epochs , table 2 ) . in figure 4 , the presence of lii @xmath26708 absorption is particularly evident from the comparison of line profile with that of m2v spectral template from pickles library .
it may be noted that a tentative detection of lii @xmath26708 in pds 11b is seen in figure 4 , when compared to the template spectrum , but can only be confirmed from spectra with better resolution and signal - to - noise .
henceforth we will be considering lii @xmath26708 absorption only in pds 11a .
@xcite also have reported lii @xmath26708 absorption line in the spectrum of pds 11a with an ew of 0.71 .
the identification of lii @xmath26708 in the spectra of pds 11a supports its t tauri membership .
we have used lithium depletion boundary ( ldb ) technique to estimate the age of pds 11a .
this method is model independent when compared to the age estimation from stellar evolutionary models and is often used for precise age estimation of young stars in moving groups @xcite . from the measured lii @xmath26708 ew it is possible to estimate the age of the t tauri star using ldb technique . since we know the spectral type of our object of interest , it is possible to set an age limit during which lii @xmath26708 absorption line is present in the spectra .
the stellar models of @xcite show that the surface lithium is depleted in m2 stars such as pds 11a ( whose effective temperature ( t@xmath36 ) is 3490 k ; * ? ? ?
* ) at an age of 15 myr .
this means that since lii @xmath26708 is present in pds 11a , the age should be less than 15 myr . to see whether the age of pds 11a matches with young stars in moving groups
, we analyzed the lii @xmath26708 ew measurements of all the known members of the nine nearby moving groups identified so far , from @xcite .
we have used those values along with the stellar effective temperature ( as given in * ? ? ?
* ) to analyze the lii @xmath26708 ew variation with temperature ( figure 7 ) . in figure 7 , pds 11a is located between 10 myr old tw hya association and 24 myr old @xmath5 pic moving group .
it may be noted there are no moving groups with lii ew measurements of stars between 10 and 24 myr .
hence , from ldb method and the analysis of lii ew distribution in moving groups , we found that pds 11a has an age of 10@xmath015 myr , which will be used in further discussion .
pds 11 is not the first case of a t tauri binary system where only one of the components show lii @xmath26708 absorption .
@xcite found that lii @xmath26708 absorption feature is seen in the secondary of the hip 112312 ( gj 871.1 a and b ) pre - ms binary system whereas it is absent in the primary .
they have used this system , which is about 12 myr old and at a distance of 24 pc , to study the ldb in pre - main sequence stars . the spectral type of gj 871.1 a and b ( m4 and m4.5 ) is closer to that of pds 11a and pds 11b ( m2 ) .
it is intriguing to see why lii @xmath26708 is present in absorption only in one component of the binary system , if they are of similar age .
we note that presence of lii @xmath26708 in absorption is not a necessary condition to classify the object as a t tauri star . @xcite
have cautioned that lithium may not always be a reliable age indicator since the lithium abundance depends on the accretion history of the star .
episodic accretion in young stars can increase the central temperature due to which lithium can get severely depleted @xcite . from our photometry
we estimated @xmath17 and @xmath37 values of pds 11a and pds 11b as [email protected] ( m@xmath38 ) , [email protected] , and [email protected] ( m@xmath39 ) , [email protected] , respectively .
the intrinsic @xmath37 color of pds 11a and pds 11b is 1.46 , considering that both are m2 stars @xcite .
our observed @xmath37 colors are found to be bluer by 0.03 mag and 0.18 mag than the intrinsic values , for pds 11a and pds 11b , respectively .
this has been noticed in previous studies and could be caused by the lower gravity of pre - ms stars with respect to the dwarfs @xcite . from the observed @xmath37 values ,
the color excess @xmath40 of both the stars is found to be negative and hence will be considered as zero from now on .
this is understandable since these are high galactic latitude objects and hence suffer little extinction .
comparison with 10@xmath015 myr isochrones from @xcite for an m2 star ( t@xmath36 = 3490 k ) indicates a mass of 0.4 m@xmath3 and luminosity in the range [email protected] l@xmath3 ( log l / l@xmath3 = @xmath01.05 @xmath01.17 ) for pds 11a .
these luminosity values imply bolometric magnitude m@xmath24 in the range of [email protected] , from which absolute @xmath17 magnitude ( m@xmath38 ) is obtained using the bolometric correction of -1.80 for m2 stars given in @xcite .
the observed v magnitude , m@xmath38 = 14.75 , and m@xmath38 indicate a distance of 114@xmath0131 pc for pds 11a .
the estimated stellar parameters are given in table 3 .
similar estimates are not possible for pds 11b as lii @xmath26708 is not present in the spectra .
lcc sp.type & [email protected] & [email protected] + t@xmath36 ( k ) & 3490 & 3490 + l@xmath24 ( l@xmath3 ) & [email protected] & . + @xmath40 & 0 & 0 + distance ( pc ) & 114@xmath0131 & .
+ age ( myr ) & 10@xmath015 & .
+ mass ( m@xmath3 ) & 0.4 & .
+ v@xmath41 ( kms@xmath27 ) & 21@xmath193@xmath42 & 10@xmath198@xmath43 + @xmath44@xmath45 ( mas yr@xmath27 ) & [email protected] & [email protected] + @xmath46@xmath45 ( mas yr@xmath27 ) & [email protected] & [email protected] + infrared excess in the energy distribution is one of the defining criteria to identify t tauri stars among the sample of low mass stars @xcite .
we made use of the archival 2mass data to estimate the near - ir @xmath47 , @xmath48 colors and use them to assess near - ir excess in pds 11a and pds 11b .
figure 8 shows the location of pds 11a & 11b in the @xmath47@xmath0@xmath48 color - color diagram . for assessing the nature of near - ir excess in pds 11 components ,
we also show ctts , wtts and transitional disk candidates in the figure .
the 32 ctts shown are of spectral types m0@xmath0m3 from the taurus star forming region , identified from @xcite and @xcite . also , included
are 14 wtts in the spectral type range m0@xmath0m3 from the taurus and chamaeleon star forming regions @xcite .
we have used the 2mass colors of a sample of 16 td candidates in taurus and chamaeleon i , taken from @xcite .
it is immediately evident from figure 8 that pds 11b shows considerable ir excess and is found to be on the ctts locus .
the @xmath49 , @xmath50 colors of the object seems to be higher than the sample of ctts used for this analysis .
however , since the location do not contain t tauri stars other than ctts , it is pretty clear that pds 11b belong to ctts category .
the near - ir excess of pds 11a is similar to the wtts / tds , suggesting significantly less hot dust material around it . in order to see whether our stars have any analogs in any of the moving group in terms of near - ir colors ,
we have represented them in @xmath47 versus @xmath48 color - color diagram .
we have included stars in the spectral range m0@xmath0m3 , from known moving groups listed in @xcite and @xcite . as seen from figure 9 ,
almost all the members of various moving groups are found to be clustered near the main sequence , similar to pds 11a .
the extreme type of ir excess seen in pds 11b is generally not seen in any other association members . in summary , both pds 11a and pds 11b are found to lie on the ctts locus , indicating the presence of warm circumstellar dust around them .
in addition to the near - ir excess , both the stars are accreting ( sect .
3.3 ) and show veiling in the observed spectra ( discussed in sect .
4.1 ) , qualifying them as classical t tauri stars .
pds 11b show very high near - ir excess and lies at the extreme end of the ctts locus .
since none of the known ctts display such high near - ir excess ( see figure 8) , it is worth exploring the nature of pds 11b from further studies . the spectral energy distribution ( sed ) of pds 11a and pds 11b is constructed with the available photometric data given in table 4 .
the seds are shown in figures 10 .
we have used bt - settl model atmospheres corresponding to the temperature ( t@xmath36 ) and gravity ( log @xmath51 ) values of of pds 11a and pds 11b . even though t@xmath36 of both the stars are 3490 k , we have taken the bt - settl atmosphere for 3500 k , which is the closest temperature for which stellar atmosphere is available .
generally , in the case of pre - ms stars , model atmospheres corresponding to log(@xmath51 ) = 4.5 is used for sed analysis .
we have verified this in the case of pds 11a , which has a log(@xmath51 ) value of 4.34 from the stellar models of @xcite . since bt - settl model atmospheres corresponding to t@xmath36 = 3490 k and log(@xmath51 ) = 4.34 is unavailable , we have used the nearest combination of t@xmath36 = 3500 k and log(@xmath51 ) = 4.5 for sed analysis .
pds 11a do not show much of ir excess whereas pds 11b shows considerable excess with the sed rising in near - ir itself ( figure 10 ) .
we found that mid - infrared magnitudes are available from @xmath52 mission for pds 11b , but the beam size includes pds 11a as well .
comparison of @xmath11 , @xmath12 , @xmath13 magnitudes of both stars from 2mass and @xcite indicate that most of the excess emission is coming from pds 11b .
pds 11b is about 2.2 mag brighter than pds 11a in @xmath53 band @xcite .
pds 11b shows considerably high @xmath47 and @xmath48 color excess ( @xmath4 1 mag ) , which is even higher than that for ctts in star forming region ( figure 8) .
we note that in addition to disk excess , the ir excess in pds 11b can also be contributed by an unseen late - type companion , which demands further investigation .
lccc this work & @xmath16 & [email protected] & [email protected] + & @xmath17 & [email protected] & [email protected] + & @xmath18 & [email protected] & [email protected] + g92 & @xmath16 & 16.27 & 16.68 + & @xmath17 & 14.76 & 15.34 + & @xmath18 & 13.70 & 14.28 + & @xmath54 & 12.55 & 13.09 + denis & @xmath54 & [email protected] & [email protected] + & @xmath11 & [email protected] & [email protected] + & @xmath13 & [email protected] & [email protected] + 2mass & @xmath11 & [email protected] & [email protected] + & @xmath12 & [email protected] & [email protected] + & @xmath13 & [email protected] & [email protected] + w95 & @xmath11 & [email protected] & [email protected] + & @xmath12 & [email protected] & [email protected] + & @xmath13 & [email protected] & [email protected] + & @xmath53 & [email protected] & [email protected] +
our analysis so far suggest that both components of the pds 11 system are of similar spectral type , m2 .
both pds 11a and pds 11b show strong h@xmath1 emission , confirming their ctts status . while pds 11b show strong excess emission in the near - ir , pds 11a show no or weak excess , indicating the that no hot dust is present close to the star .
intriguingly , pds 11a show lithium absorption , suggesting that it is 10@xmath015 myr old ; pds 11b do not show lithium absorption .
below we discuss more aspects on veiling and binarity of this system .
the presence of excess continuum emission , referred to as veiling , is often observed in classical t tauri stars @xcite . in the magnetospheric accretion model ,
veiling is due to the dissipation of energy in the post - shock region at the base of the magnetic funnel , close to the stellar surface @xcite .
@xcite suggested a method to estimate veiling in t tauri stars .
they found that measured ew values of cai @xmath24227 absorption line follow a relation with spectral type , of the form ew(cai ) = @xmath0189.218 + 7.36@xmath55 @xmath0 0.072@xmath56 .
since both pds 11a and pds 11b are of m2 spectral type , ` @xmath55 ' corresponds to 60 @xcite .
hence the expected cai @xmath24227 ew should be 6.8 , if veiling is not present .
in the case of pds 11a , ew(cai ) measured on 2016 february 13 agrees with the expected value , whereas it changed to [email protected] , when measured from the spectra taken on 2016 march 15 . during the first epoch ,
the measured ew value of cai @xmath24227 is almost similar to that expected if veiling is not present .
but in the second epoch , the line gets veiled , suggesting that emission from accretion continuum is filling - in the absorption line .
the situation is even more interesting in the case of pds 11b which show ew(cai ) value of [email protected] from the spectra taken on 2016 february 13 and the line is almost not visible in the spectra taken on 2016 march 15 . since the measured ew of cai @xmath24227 is considerably lower than the empirical estimates , pds 11b must be undergoing strong veiling .
we estimated the effect of veiling on spectral type determination for both the stars using the ew(cai ) relation given in @xcite .
we found that the spectral type of pds 11a could change from m2 to m1 for a change in ew(cai ) from 6.8 to 5.6 . in the case of pds 11b ,
the cai @xmath24227 line is almost completely veiled and hence the spectral type can shift to earlier type by @xmath45 subtypes . the fact that the spectra of both the stars in pds 11 are veiled has important implications .
firstly , this supports our result that pds 11a and pds 11b are ctts in active accretion phase .
secondly , this explains the absence of lii @xmath26708 absorption line in pds 11b .
it is quite possible that pds 11b is of the same age as pds 11a and lii @xmath26708 is present , but the line is filled in because of strong veiling .
we examined this aspect by comparing the lithium abundance values listed in @xcite with the shift in spectral type due to veiling . in the previous paragraph
, we described that veiling was present in the spectrum of pds 11b obtained on 2016 february 13 , while pds 11a did not show evidence for veiling during that epoch .
the equivalent width of cai @xmath24227 in pds 11b was found to be 6 , whereas the expected value in the absence of veiling is 6.8 .
this reduction in absorption strength due to veiling will shift the spectral type from m2 to m1 for pds 11b , according to the ew(cai ) relation given in @xcite .
thus , based on ew(cai ) , the spectral type of pds 11b was of m1 and pds 11a of m2 type , for the epoch 2016 february 13 .
the age of pds 11a is estimated in the range 10@xmath015 myr . from @xcite
stellar models it is seen that at an age of 10 myr , the ratio of surface lithium abundance to initial abundance is @xmath427 times higher in an m2 star like pds 11a , when compared to m1 star like pds 11b . since we found lii @xmath26708 absorption in pds 11a to be 0.43 , the ew(lii ) expected for pds 11b is around 0.02 .
this value is far below the detection limit of our instrument , which supports our proposition that lii @xmath26708 might be present in pds 11b , but not visible in the spectrum due to strong veiling .
we have used the banyan ii webtool to check whether our candidates are associated with any of the nearby kinematic groups .
banyan ii is a bayesian analysis tool which makes use of the position and space velocities of the object to assess the match with the database of nearby ( @xmath28100 pc ) moving groups , younger than 100 myr @xcite .
the heliocentric radial velocity and proper motion of pds 11a and pds 11b are taken from the literature and are given in table 3 .
from the analysis we found that pds 11a and pds 11b are not associated with any known moving group .
banyan ii analysis gives 100% probability that pds 11a and pds 11b belong to young field population @xcite .
it is quite possible that our object parameters may not match with that of any known association and hence has been ascribed to young field population .
although the pds 11 system has been treated as binary in the literature @xcite , it is yet to be demonstrated that pds 11a & 11b are gravitationally bound .
it has been identified as visual binaries in washington double star catalog @xcite , but that do not guarantee them being gravitationally bound .
instead , we found that the proper motion in ra and dec for both the stars are similar within uncertainties ( table 3 ) .
this suggests that pds 11a and pds 11b form a binary system .
considering pds 11 at a distance of 114@xmath0131 pc , with a separation of 8.8@xmath7 between the components @xcite , the physical separation between pds 11a and pds 11b is @xmath4 1003@xmath01153 au .
the proximity of two ctts at such a separation argues against chance alignment and suggests that they are part of a bound system .
hence , the age and distance estimated for pds 11a can as well be applied to pds 11b . from a study of t tauri stars in taurus - auriga star - forming region , @xcite obtained a relation to estimate the lifetime of the disk in t tauri stars ( @xmath57 ) , @xmath57 = 4@xmath1410@xmath58(m / m@xmath3)@xmath59 , in terms of the mass of the parent star ( m ) . since the mass of pds 11a and pds 11b are estimated to be 0.4 m@xmath3 , the disk lifetime is around 2.0 myr .
evidently it is quite puzzling how a disk which harbors enough gas and dust survive in 10@xmath015 myr old system like pds 11 .
t tauri binary systems with disks at ages older than the typical disk dissipation timescales have been reported in the literature .
most of them belong to nearby ( @xmath28 100 pc ) young moving groups .
they are : 8 myr old binary systems tw hya @xcite , hr 4796 @xcite , twa 30 @xcite , t cha @xcite , 20 myr old v4046 sgr @xcite , and lds 5606 @xcite .
these objects belong to class of wide binaries , where the separation between the components is in the range 1.7 kau ( for lds 5606 , * ? ? ?
* ) to 41 kau ( for tw hya@xmath0twa 28 system , * ? ? ?
* ) . among them , only twa 30 and lds 5606 are t tauri binary systems in which both components are accreting .
the binary components of twa 30 system are separated by @xmath43400 au , show nearly edge - on orientation and are of similar spectral type , m5 and m4 , respectively @xcite .
lds 5606 is an m5/m5 t tauri binary system at a separation of @xmath41700 au and they are members of @xmath5 pic moving group @xcite .
to summarize , pds 11 is the third such system , after twa 30 and lds 5606 , which belong to the interesting class of old , dusty , wide binary classical t tauri systems in which both components undergo active accretion .
it is quite possible that pds 11a is a transition disk candidate since it is accreting and lacks hot dust material close to the star ( see sect .
3.3 & 3.6 ) .
however , further studies are needed to confirm other transition disk properties such as mid- and far - ir excess and the presence of outer disk in pds 11a .
also , further observations are needed to assess whether the near - ir excess in pds 11b is entirely due to circumstellar material or due to the contribution from late - type companion .
if confirmed , this would be the first known example of a @xmath6010 myr old binary system , where one of the components harbor a radially continuous full disk , while the other is surrounded by a disk with inner hole or gap .
we have analyzed the star / disk properties and derived the spectral type of the t tauri binary system pds 11 from optical photometry , spectroscopy and infrared spectroscopic observations .
our analysis indicates that pds 11 is the new addition , after twa 30 and lds 5606 , to the interesting class of old , dusty , wide binary classical t tauri systems in which both components are actively accreting .
the main conclusions from this study are listed below . *
the spectral type of pds 11a and pds 11b were not known .
we have classified both as m2-type making use of the tio @xmath27050 band feature , with the aid of tio5 index and the relations from @xcite .
* pds 11a and pds 11b are found to have h@xmath1 emission strength of @xmath425 , which is higher than the threshold value of chromospherically active stars .
the median accretion rate derived from h@xmath1 emission line is around @xmath4 1.0@xmath1410@xmath26 m@xmath3 yr@xmath27 for pds 11a .
pds 11b show very high near- and mid - infrared excess . the emission lines of [ oi ] @xmath26300 and hei @xmath25876 , indicative of accretion process , are present in the spectrum of pds 11b .
all these evidences conclusively prove that pds 11a and pds 11b are classical t tauri stars .
it needs to be assessed from future studies how an active accretion disk sustains in a 10@xmath015 myr old system like pds 11a .
* we found that pds 11a is less than 15 myr from age dating using lithium depletion boundary method .
further , from the comparison of lii @xmath26708 ew with that of young stars in moving groups , the age is constrained in the range 10@xmath015 myr . *
since lii @xmath26708 is not present in the spectrum of pds 11b , the stellar parameters other than spectral type and temperature were not determined .
however , since we prove that pds 11a and pds 11b form a binary system , age and distance of pds 11b is taken to be similar to pds 11a . * from our analysis
pds 11 is identified as a binary system with component masses of 0.4 m@xmath3 , luminosity of [email protected] l@xmath3 and at a distance of 114@xmath0131 pc . + * from the analysis with banyan ii webtool , pds 11a and pds 11b were not identified as members of any known moving group and hence is considered as a young field binary system .
we would like to thank the staff at iao , hanle and its remote control station at crest , hosakote for their help during the observation runs .
this research uses the simbad astronomical data base service operated at cds , strasbourg .
this publication made use data of 2mass , which is a joint project of university of massachusetts and the infrared processing and analysis centre / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
appenzeller , i. , mundt , r. 1989 , a&ar , 1 , 291 baraffe , i. , homeier , d. , allard , f. , et al .
2015 , , 577 , a42 baraffe , i. , chabrier , g. 2010 , , 521 , a44 barrado y navascus , d. , martn , e. l. 2003 , , 126 , 2997 bell , c. p. m. , mamajek , e. e. , naylor , t. 2015 , , 454 , 593 bertout , c. , siess , l. , cabrit , s. 2007 , , 473 , l21 bertout , c. 1989 , , 27 , 351 bessell , m. s. 1990 , , 102 , 1181 binks , a. s. , jeffries , r. d. 2014 , , 438 , l11 bodenheimer , p. 1965
, , 142 , 451 calvet , n. , muzerolle , j. , briceno , c. , et al .
2004 , , 128 , 1294 calvet , n. , & gullbring , e. 1998 , , 509 , 802 carpenter , j. m. 2001 , , 121 , 2851 chabrier , g. , baraffe , i. , plez , b. 1996 , , 459 , l91 cutri r. m. , et al .
2003 , vizier online data catalog , 2246 , 0 da silva , l. , torres , c. a. o. , de la reza , r. , et al .
2009 , , 508 , 833 de la reza , r. , torres , c. a. o. , quast , g. , et al .
1989 , , 343 , l61 elliott , p. , bayo , a. , melo , c. h. f. , et al .
2016 , , 590 , a13 fang , m. , van boekel , r. , wang , w. , et al .
2009 , , 504 , 461 furlan , e. , luhman , k. l. , espaillat , c. , et al .
2011 , , 195 , 3 furlan , e. , hartmann , l. , calvet , n. , et al .
2006 , , 165 , 568 gagn , j. , lafrenire , d. , doyon , r. , et al .
2014 , , 783 , 121 gregorio - hetem , j. , lepine , j. r. d. , quast , g. r. , et al .
1992 , , 103 , 549 hamann , f. , & persson , s. e. 1992 , , 82 , 247 hartigan , p. , edwards , s. , ghandour , l. 1995 , , 452 , 736 hartmann , l. 2009 , accretion processes in star formation : second edition ( cambridge university press ) hartmann , l. , hewett , r. , calvet , n. 1994 , , 426 , 669 herbig g. h. , bell k. r. 1988 , third catalog of emission - line stars of the orion population , lick observatory , santa cruz herbig , g. h. 1962 , adv .
herczeg , g. j. , & hillenbrand , l. a. 2014 , , 786 , 97 herczeg , g. j. , & hillenbrand , l. a. 2008 , , 681 , 594 ingleby , l. , calvet , n. , herczeg , g. , et al .
2013 , , 767 , 112 johns - krull , c. m. , valenti , j. a. 2001 , , 561 , 1060 joy , a. h. 1949 , , 110 , 424 joy , a. h. 1945 , , 102 , 168 kastner , j. h. , thompson , e. a. , montez , r. 2012 , , 747 , l23 kastner , j. h. , sacco , g. g. , montez , r. , et al . 2011 , , 740 , l17 kastner , j. h. , zuckerman , b. , bessell , m. 2008 , , 491 , 829 kastner , j. h. , zuckerman , b. , weintraub , d. a. , et al .
1997 , sci , 277 , 67 kenyon , s. j. , hartmann , l. 1995 , , 101 , 117 kim , k. h. , watson , d. m. , manoj , p. et al .
2016 , preprint , ( arxiv:1604.07907 ) kim , k. h. , watson , d. m. , manoj , p. et al .
2013 , , 769 , 149 koenigl , a. 1991 , , 370 , l39 koornneef , j. 1983 , , 128 , 84 looper , d. l. , mohanty , s. , bochanski , j. j. , et al . 2010a , , 714 , 45 looper , d. l. , bochanski , j. j. , burgasser , a. j. , et al .
2010b , , 140 , 1486 malo , l. , doyon , r. , lafrenire , d. , et al .
2013 , , 762 , 88 mamajek , e. e. 2016 , in iau , symp 10 , 314 , 21 manara , c. f. , beccari , g. , da rio , n. , et al .
2013 , , 558 , a114 manoj , p. , kim , k. h. , furlan , e. , et al .
2011 , , 193 , 11 martn , e. l. 1998 , , 115 , 351 mason , b. d. , wycoff , g. l. , hartkopf , w. i. , et al .
2001 , , 122 , 3466 mcgehee , p. m. 2013 , handbook of star forming regions , volume ii : the southern sky asp monograph publications , vol . 5 .
edited by bo reipurth , p.813 meyer , m. r. , calvet , n. , hillenbrand , l. a. 1997 , , 114 , 288 muzerolle , j. , calvet , n. , hartmann , l. 1998 , , 492 , 743 natta , a. , testi , l. , randich , s. 2006 , , 452 , 245 ninan j. p. et al . , 2014 , j. astron .
instrum . , 3 , 1450006 pecaut , m. j. , & mamajek , e. e. 2013 , , 208 , 9 pickles , a. j. 1998 , , 110 , 863 principe , d. a. , sacco , g. , kastner , j. h. , et al .
2016 , , 459 , 2097 qi , z. , yu , y. , bucciarelli , b. , et al . 2015 , , 150 , 137 reid , i. n. , hawley , s. l. , gizis , j. e. 1995 , , 110 , 1838 rieke , g. h. , & lebofsky , m. j. 1985 , , 288 , 618 rodriguez , d. r. , zuckerman , b. , faherty , j. k. , et al .
2014 , , 567 , a20 rucinski , s. m. , krautter , j. 1983 , , 121 , 217 sergison , d. j. , mayne , n. j. , naylor , t. , et al .
2013 , , 434 , 966 slesnick , c. l. , carpenter , j. m. , hillenbrand , l. a. , et al .
2006 , , 132 , 2665 soderblom , d. r. 2010 , , 48 , 581 song , i. , bessell , m. s. , zuckerman , b. 2002 , , 581 , l43 tachihara , k. , neuhuser , r. , fukui , y. 2009 , , 61 , 585 teixeira , r. , ducourant , c. , chauvin , g. , et al .
2008 , , 489 , 825 torres , c. a. o. , quast , g. r. , melo , c. h. f. , sterzik , m. f. 2008 , handbook of star forming regions , volume ii : the southern sky asp monograph publications , vol . 5 . edited by bo reipurth , p.757 white , r. j. , & basri , g. 2003 , , 582 , 1109 whitelock , p. , menzies , j. , feast , m. , et al .
1995 , , 276 , 219 zuckerman , b. , vican , l. , rodriguez , d. r. 2014 , , 788 , 102 zuckerman , b. , & song , i. 2004 , , 42 , 685 | we present results of our study of pds 11 binary system , which belongs to a rare class of isolated , high galactic latitude t tauri stars .
our spectroscopic analysis reveals that pds 11 is a m2@xmath0m2 binary system with both components showing similar h@xmath1 emission strength .
both the components appear to be accreting , and are classical t tauri stars .
the lithium doublet lii @xmath26708 , a signature of youth , is present in the spectrum of pds 11a , but not in pds 11b . from the application of lithium depletion boundary age - dating method and a comparison with the lii @xmath26708 equivalent width distribution of moving groups , we estimated an age of 10@xmath015 myr for pds 11a .
comparison with pre - main sequence evolutionary models indicates that pds 11a is a 0.4 m@xmath3 t tauri star at a distance of 114@xmath0131 pc .
pds 11 system does not appear to be associated with any known star forming regions or moving groups .
pds 11 is a new addition , after twa 30 and lds 5606 , to the interesting class of old , dusty , wide binary classical t tauri systems in which both components are actively accreting . |
the _ global _ treatment of asymptotically flat ( or , more generally , asymptotically simple ) spacetimes as models for isolated systems is an important topic , both in mathematical and in numerical relativity .
in particular , this is relevant for a rigorous investigation of gravitational wave signals , which can be uniquely read off only at null infinity @xmath0 .
indeed , it has been shown that certain tail decay rates at timelike boundaries ( which are often introduced as `` artificial '' outer boundaries for numerical computations ) and at null infinity are different , so that an appropriate treatment of infinity is crucial , see , e.g. , @xcite and references therein .
an important ingredient for such a rigorous treatment is penrose s conformal boundary @xcite , which is obtained by isometrically embedding the spacetimes into larger lorentzian manifolds and conformally rescaling the metric , i.e. the `` physical '' metric is replaced by a conformal metric and a conformal factor . a formulation of einstein s field equations adapted to this setting , the conformal field equations ( cfe ) , was given by friedrich @xcite . in the context of this approach , it is well - known that certain mathematical difficulties arise in the region near spacelike infinity @xmath1 . in order to cope with this problem ,
friedrich has introduced yet another formulation of the field equations , the generalised cfe @xcite . moreover ,
a novel representation of spacelike infinity has been given : @xmath1 , in regular cases thought of before as a point , is `` blown up '' to a cylinder @xmath2 $ ] , which connects past and future null infinity @xmath3 .
the cylinder - formulation is very appealing for numerical calculations as well .
recently , the `` core '' equations of the generalized cfe , a spin-2 system on minkowski background describing linearised gravitational fields near spacelike infinity , have been solved numerically with a finite - difference method ( in particular , using fourth - order runge - kutta in time ) , see @xcite . a typical drawback in this approach ( and other finite - difference schemes )
is the restriction of the time step by the courant - friedrichs - lewy ( cfl ) condition . in particular , as one approaches @xmath0 , the characteristic speeds of the system diverge .
consequently , the requirement that the numerical domain of dependence contain the analytical one enforces smaller and smaller time steps .
hence , it is only possible to come close to @xmath0 , using a very large number of grid points in the time direction , without actually reaching it .
however , since gravitational wave signals are defined at @xmath0 itself , it is desirable to include this boundary into the numerical domain . in the present article
, we address this problem in a simplified setting : we consider the conformally invariant wave equation on a minkowski background .
moreover , we restrict ourselves to the @xmath4 dimensional problem in spherical symmetry .
the resulting wave equations are simpler than the above mentioned spin-2 system , but mirror already important mathematical properties and difficulties of the general problem .
hence , a successful numerical treatment of this model will lead the way to future applications to the linearised , or even to the full einstein equations .
our main objective is to demonstrate that the wave equation near the cylinder at spacelike infinity ( and even in the entire minkowski space ) can be solved with a fully pseudospectral scheme .
the underlying numerical method , which is based on the approach by ansorg _
et al . _
@xcite for solving elliptic pdes , was introduced in @xcite and later applied to several physical problems in @xcite .
the main idea is to expand the unknown functions in terms of chebyshev polynomials , both with respect to space _ and _ time .
this provides a highly implicit method for which , in particular , no cfl condition is required , since the solution is simultaneously obtained in the entire numerical domain .
consequently , we will see that we can reach @xmath5 with a small number of grid points ( about 20 - 30 points in time direction ) instead of only approaching it asymptotically .
in the following , we solve _ cauchy problems _ with data on a hypersurface @xmath6 and evolve them to future null infinity @xmath5 .
thereby , we first concentrate on a vicinity of @xmath1 only , and afterwards study the same problem for all values of a radial coordinate @xmath7 , i.e. in the entire minkowski space and not only close to infinity .
a different type of initial value problem that has frequently been studied in general relativity and that is again gaining more attention in recent times , is an _ ( asymptotic ) characteristic initial value problem with data on past null infinity _ @xmath8 .
we refer to @xcite for some interesting mathematical results . here
, we demonstrate that this type of initial value problem , applied to the conformally invariant wave equation , can also be solved with the fully pseudospectral method .
the different initial value problems considered here are illustrated in fig .
[ fig : ivps ] below .
this paper is organized as follows .
we begin with a quick overview of the conformally invariant wave equation and the fully pseudospectral scheme in sec .
[ sec : background ] .
afterwards , in sec .
[ sec : i0 ] , we apply this method to the wave equation near spacelike infinity . in sec .
[ sec : all ] , we consider different coordinates that cover all of minkowski space and again solve the wave equation . finally , in sec .
[ sec : discuss ] , we summarize our results .
in @xmath9 spacetime dimensions , the conformally invariant wave equation is @xmath10 where @xmath11 is the ( inverse ) metric and @xmath12 the corresponding ricci scalar .
this equation is conformally invariant for @xmath13 with conformal weight @xmath14 , i.e. if @xmath15 solves the equation for a metric @xmath16 , then @xmath17 solves the equation for a conformally related metric @xmath18 with conformal factor @xmath19 . here
we consider the equation in minkowski space , i.e. we assume that the physical metric @xmath20 is the minkowski metric , and we restrict ourselves to @xmath21 dimensions , where we have the equation @xmath22 then the conformal weight is @xmath23 .
moreover , we consider only spherically symmetric solutions @xmath24 , which depend on time @xmath25 and on a radial coordinate @xmath7 alone . in this article
we look at cauchy problems , where the function values and first time derivatives of @xmath13 can be prescribed at the initial cauchy surface .
we also study characteristic initial value problems , where only the initial function values can be specified . in the following
we give a short summary of the fully pseudospectral scheme , which will enable us to find highly accurate numerical solutions to the conformally invariant wave equation . for more details
we refer to @xcite .
the basic idea is to express the unknown function(s ) in terms of truncated series of chebyshev polynomials @xmath26 , both with respect to space and time , and to reduce the differential equation ( in our case the conformally invariant wave equation ) to an algebraic system of equations . to this end
, the physical domain is first mapped to a unit square by introducing spectral coordinates @xmath27\times[0,1]$ ] such that the initial surface corresponds to @xmath28 .
then we enforce the initial conditions by expressing the unknown function @xmath29 in terms of another unknown @xmath30 with an expansion @xmath31 depending on the type of initial data ( cauchy problem or characteristic initial value problem ) .
the functions @xmath32 and @xmath33 can be obtained from the given initial data at @xmath28 . in a next step ,
we approximate the new unknown @xmath34 via @xmath35 for given spectral resolution ( number of polynomials ) @xmath36 and @xmath37 . from this approximation for @xmath34
we can then easily find approximations for @xmath13 and its derivatives .
an algebraic system of equations for the chebyshev coefficients @xmath38 can be obtained by requiring that the wave equation ( and suitable boundary or regularity conditions , if applicable ) are satisfied at a set of collocation points . here , we choose _ gauss - lobatto _ nodes , which are defined by @xmath39 note that , instead of considering the resulting algebraic system of equations as a system for @xmath38 , we actually solve directly for the function values @xmath40 at the nodes . finally , starting from some initial guess ( in our examples we can choose the trivial solution @xmath41 for all @xmath42 ) , we solve this system iteratively with the _ newton - raphson _ method .
in order to describe a vicinity of spacelike infinity @xmath1 and to resolve the cylinder - like structure of @xmath1 , we first introduce suitable coordinates . to this end , we start from the physical metric @xmath16 of minkowski space , which takes the form @xmath43 in spherical coordinates @xmath44 .
then we define new coordinates @xmath25 and @xmath7 as @xmath45 the metric can now be expressed as @xmath46 in terms of the conformal metric @xmath47 with the conformal factor @xmath48 note that this is a special case of the more general coordinates considered by friedrich @xcite , corresponding to the choice @xmath49 for the free function @xmath50 that appears in friedrich s transformation .
illustration of the two types of cauchy problems ( a , b ) and the two types of characteristic initial value problems ( c , d ) under consideration .
initial data are prescribed at the thick lines and evolved into the shaded regions .
spatial infinity @xmath1 is indicated with dashed lines , and dotted lines show the positions of null infinity @xmath5 and @xmath8 .
the coordinates in ( a ) are adopted for studying a vicinity of @xmath1 ( see sec . [ sec : i0 ] ) , whereas the coordinates in ( b - d ) cover the entire minkowski space ( see sec . [ sec : all ] ) . ]
now we can look at the conformally invariant wave equation corresponding to the metric .
since the ricci scalar @xmath12 of this metric vanishes identically , we only have to consider the first term in .
this leads to @xmath51 where a comma denotes partial derivatives . in our setting
, the coordinates are defined in the domain @xmath52 $ ] , @xmath53 .
the boundary @xmath54 , @xmath55 corresponds to null infinity @xmath3 , and @xmath1 is located at @xmath56 , see fig .
[ fig : ivps]a .
the characteristics of the wave equation are lightlike curves with respect to the metric @xmath57 and have the form @xmath58 where different values of the constant @xmath59 correspond to different characteristics .
we intend to solve a cauchy problem with data at @xmath6 , which will then be evolved until @xmath60 , and we would like to obtain the solution @xmath13 in some vicinity of @xmath1 , say for @xmath61 . on the other hand , if we strictly restrict ourselves to the domain @xmath62\times[0,1]$ ] , then we would need to prescribe boundary conditions at the outer boundary @xmath63 .
however , since there is no physical reason why @xmath13 should satisfy any particular conditions at @xmath63 , `` artificial '' conditions at this boundary even though mathematically perfectly valid would be rather unphysical .
therefore , we instead choose a slightly larger domain that contains the square @xmath64\times[0,1]$ ] , but where the outer boundary is the ingoing characteristic @xmath65 , see the shaded region in fig .
[ fig : ivps]a .
then there is no need for boundary conditions at this curve .
instead , we can impose the wave equation itself there . at @xmath56 (
i.e. at @xmath1 ) the wave equation becomes an _
intrinsic _ equation , @xmath66 in accordance with the observation that the cylinder at spacelike infinity is a _ totally characteristic surface _ also for friedrich s conformal field equations .
the latter equation has the first integral @xmath67 , @xmath68 .
this shows that @xmath13 will be singular at @xmath56 , @xmath60 , i.e. the point where @xmath1 and @xmath5 intersect , also is a well - known property of friedrich s conformal field equations , see @xcite .
but as we see , this problem already shows up in the wave equation discussed here . in both cases ,
the requirement of regularity leads to regularity conditions in the form of restrictions of the initial data . ] , unless we choose initial data at @xmath6 with @xmath69 at @xmath56 ( for which @xmath70 ) . in the following ,
we restrict ourselves to this choice , in which case @xmath13 will be constant at @xmath1 .
therefore , for our numerical method , we can choose @xmath69 as a regularity condition there .
another condition at @xmath1 follows from first differentiating with respect to @xmath7 and then performing the limit @xmath71 .
this leads to @xmath72 , i.e. @xmath73 at @xmath56 .
we will use this condition ( instead of the condition @xmath69 ) at the single point$ ] with some @xmath74 , then the condition @xmath69 is sufficient at @xmath75 , @xmath56 . for @xmath76 , however , it turns out that the alternative condition needs to be imposed in order to guarantee numerical stability . ] @xmath60 , @xmath56 in the form @xmath77 where the right hand side is obtained from the given values for @xmath78 at @xmath6 .
we can easily derive the general solution to the wave equation , which is useful for testing the accuracy of our numerical method .
starting from the solution @xmath79/\tilde r$ ] of the wave equation with respect to the physical metric @xmath16 , where @xmath80 and @xmath81 are two free functions at the origin @xmath82 of minkowski space , then we need to choose @xmath83 .
however , since we are currently only interested in a vicinity of spatial infinity , we can choose @xmath80 and @xmath81 independently .
] we obtain ( cf .
[ sec : confwe ] ) @xmath84 in terms of free functions @xmath85 and @xmath86 ( related to @xmath80 and @xmath81 via @xmath87 and @xmath88 ) .
we apply the fully pseudospectral scheme as described in sec .
[ sec : spectral ] . for that purpose
, we introduce spectral coordinates @xmath89 $ ] via @xmath90 the outer boundary ( the ingoing characteristic ) is then given by @xmath91 , whereas @xmath92 corresponds to @xmath1 .
as initial data we prescribe the values of @xmath13 and @xmath78 at @xmath93 , which is numerically implemented with an expansion of the form . as discussed above , we impose the boundary condition @xmath69 at @xmath94 , with exception of the point @xmath92 , @xmath95 , where we use condition .
another exceptional point is @xmath96 on the initial surface , where @xmath69 is automatically satisfied , because , as discussed above , we choose initial data subject to this condition . at this point
, we instead use the condition @xmath97 ( which is a simple consequence of @xmath69 at @xmath56 ) . at all other grid points
we impose the conformally invariant wave equation . in order to test the numerical accuracy of our method ,
we choose three different exact solutions by specifying the free functions @xmath85 and @xmath86 in the general solution as given in table [ tab : exa ] .
.[tab : exa ] the functions @xmath98 and @xmath99 for three example solutions to the conformally invariant wave equation near spatial infinity , cf .
[ cols="<,<,<",options="header " , ] first we consider the cauchy problem ( i ) .
here we prescribe the values of @xmath13 and @xmath78 at @xmath6
. then we solve the wave equation with the following boundary conditions : at @xmath56 we impose @xmath100 , with exception of the point @xmath101 , where we use the condition ; at @xmath1 we impose @xmath69 , but at the intersection of @xmath1 with @xmath6 , where this condition is already satisfied by our choice of initial data , we have the condition @xmath97 .
similarly to our numerical studies near @xmath1 , we solve the problems with different resolutions and compare the numerical error to the interpolation error .
the resulting convergence plots are shown in fig .
[ fig : conb ] . convergence plots for the cauchy problem from fig .
[ fig : ivps]b .
the three examples are described in table [ tab : exb ] . ]
we observe a similar behaviour as for the wave equation near @xmath1 : the convergence curves are initially approximately linear , corresponding to exponential convergence , until saturation near machine accuracy is reached .
moreover , the error due to the numerics is about 1 - 3 orders of magnitude larger than the interpolation error . in the case of example 2 , we observe that the numerical error is again growing for @xmath102 .
this happens occasionally in the fully pseudospectral scheme due to accumulation of rounding errors for higher resolutions .
next we consider the characteristic problems ( ii ) and ( iii ) , where , in the latter case , we both solve the regular and singular wave equations . for problem ( ii ) , we prescribe the values of @xmath13 at @xmath8
. then we solve the wave equation with the boundary condition @xmath100 at @xmath56 and with condition at the intersection of @xmath56 and @xmath8 . in case of the characteristic problem ( iii ) for the regular wave equation
, we again prescribe the values of @xmath13 at @xmath8 . however , this time this initial condition is not achieved with an expansion of the form it turns out that this leads to numerical instabilities in the present case where the entire minkowski space is considered .
instead , we just use the condition that @xmath13 be equal to the prescribed initial values at @xmath28 , i.e. here the function values of @xmath13 themselves ( and not @xmath34 ) are the unknowns in the numerical method . furthermore , at @xmath56 we again impose the regularity condition @xmath100 .
finally , in the characteristic problem ( iii ) for the singular wave equation , we prescribe values for @xmath78 at @xmath8 .
as in the regular case before , we impose this condition directly instead of enforcing it with an expansion . as boundary conditions we use @xmath100 at @xmath56 and @xmath103 at @xmath5 and @xmath1 .
convergence plots for the three described problems , using the same exact solutions from table [ tab : exb ] as before , are shown in fig .
[ fig : conc ] .
( this time we do not compare the numerical error with the interpolation error , which would lead to similar plots as above . ) convergence plots for the characteristic initial value problems .
again we have studied the three numerical examples described in table [ tab : exb ] .
initial data are prescribed at @xmath8 and then evolved until @xmath6 ( see fig .
[ fig : ivps]c ) or to @xmath5 ( fig .
[ fig : ivps]d ) . in the latter case ,
both the regular wave equation for the conformal metric and the singular wave equation corresponding to the physical metric are solved . ]
as one should expect , we observe the smallest errors in the case where only the lower half @xmath104 of minkowski space is considered , whereas the accuracy drops if we evolve the data on @xmath8 to @xmath5 , since the solution has then to be computed on a much larger domain .
interestingly , in the latter case , we obtain comparable errors to calculate the conformally related function @xmath105 , then we would loose some accuracy .
for , applying lhpital s rule , we would need to differentiate @xmath15 at the boundaries ( where @xmath106 ) . and with each differentiation of a function approximated in terms of chebyshev polynomials , the error increases by about 1 order of magnitude . ] for the regular and singular formulations the only difference being that for the singular case , slightly higher resolutions are needed to reach the saturation level .
this indicates that , from a numerical point of view , the regular formulation of an equation is not necessarily preferred to a singular formulation provided suitable numerical techniques are applied to treat the singular case .
an example for the distribution of the error over the numerical domain is shown in fig . [ fig : errb ] ( which should be compared to the similar plot in fig .
[ fig : erra ] ) .
this time the largest error is assumed at or near @xmath92 , corresponding to the origin @xmath56 of minkowski space . also note that the error is not zero at @xmath28 , since for this calculation the initial conditions are not enforced exactly , as described above . in some numerical examples , a peak in the error
is found at @xmath92 , @xmath95 , but otherwise we obtain very similar plots for other examples and resolutions . the numerical error for the ( regular ) wave equation in the entire minkowski space .
the plot is obtained for example solution 2 from table [ tab : exb ] with spectral resolution @xmath107 . ]
as a model problem for solving einstein s equations near the cylinder at spacelike infinity @xmath1 , we have considered the conformally invariant wave equation on minkowski background . with the fully pseudospectral scheme from @xcite , it was possible to obtain highly accurate numerical solutions ( with numerical error in the order @xmath108 to @xmath109 , close to machine accuracy of our double precision code ) for a small number of grid points .
in particular , the method does not fail at @xmath0 or at the intersection @xmath110 of the cylinder and @xmath3 , where the equation degenerates . instead , it was possible to include both @xmath0 and @xmath110 into the numerical domain .
this is in contrast to finite - difference approaches , where @xmath0 can only be reached asymptotically as a consequence of the cfl condition .
we have also compared the numerical treatment of the regular wave equation ( with respect to the conformal metric ) to the singular wave equation ( with respect to the `` physical '' metric ) . in both cases we have achieved comparable numerical accuracy , which shows that both formulations have equal rights from a numerical point of view ( whereas the `` regular formulation '' is preferential in analytical approaches , since formally singular terms complicate the calculations ) .
these observations suggest that our fully pseudospectral scheme should be a promising method also for treating more general equations ( e.g. , the spin-2 system or the full einstein equations ) and less symmetric spacetimes ( e.g. , axisymmetric dimensional problems @xcite . ] or even completely non - symmetric solutions ) .
cruz - osorio , a. , gonzlez - jurez , a. , guzmn , f. s. , lora - clavijo , f. d. , _ numerical solution of the wave equation on particular space - times using cmc slices and scri - fixing conformal compactification _
f ' is . *
56 * , 456 ( 2010 ) hennig , j. and ansorg , m. , _ a fully pseudospectral scheme for solving singular hyperbolic equations on conformally compactified space - times _ , journal of hyperbolic differential equations * 6 * , 161 ( 2009 ) | we study the scalar , conformally invariant wave equation on a four - dimensional minkowski background in spherical symmetry , using a fully pseudospectral numerical scheme .
thereby , our main interest is in a suitable treatment of spatial infinity , which is represented as a cylinder .
we consider both cauchy problems , where we evolve data from a cauchy surface to future null infinity , as well as characteristic initial value problems with data at past null infinity , and demonstrate that highly accurate numerical solutions can be obtained for a small number of grid points . |
the problem of a random walk on a two - dimensional lattice with a single interior source point and absorbing boundaries was first considered by courant _
_ @xcite in 1928 where general properties of the solution were discussed .
this problem was solved exactly in 1940 @xcite for the case of random walks on a square lattice with rectangular absorbing boundaries .
the problem on a triangular lattice with finite absorbing boundaries was considered in 1963 @xcite where the exact solution was given for an approximation of the problem using straight boundaries rather than the true zig - zag boundaries of the triangular lattice .
indeed the authors of this paper remarked that `` an explicit solution of the difference equation can hardly be obtained if these boundary conditions are used . '' other variants of the problem on the square lattice have been solved exactly @xcite however the problem on the triangular lattice with true zig - zag triangular lattice boundaries has remained unsolved . in this paper
we give the exact solution for this problem .
the problem of random walks on finite lattices with absorbing boundaries is fundamental to the theory of stochastic processes @xcite and has numerous applications .
these include potential theory @xcite , electrical networks @xcite , surface diffusion @xcite and diffusion - limited aggregation ( dla ) @xcite .
for example , the exact results for the square lattice problem @xcite have been used to expedite the growth of large dla clusters on the square lattice @xcite and the formulae derived in this paper could similarly be used on the triangular lattice .
a schematic illustration showing random walk pathways ( dashed lines ) on an equi- angular triangular lattice is shown in fig .
1 . a natural co - ordinate system for the triangular lattice is to label the lattice vertices by the intersection points @xmath0 of horizontal straight lines @xmath1 and slanted straight lines @xmath2 parallel to the @xmath3 axis shown in fig .
1 . in this co - ordinate system
the expectation that a random walk on the lattice visits a site @xmath0 , before exiting at a finite boundary , is given by the non - separable difference equation @xmath4.\end{aligned}\ ] ] in this paper we adopt a different co - ordinate system , which may at first appear less natural but it has the advantage that the governing equations are separable .
in the co - ordinate system used here we label the vertices of the triangular lattice by the intersection points @xmath0 of horizontal straight lines @xmath1 and vertical zig - zag lines @xmath2 ( see fig .
2 ) . in the absence of any source terms
the expectation that a random walker visits a lattice site @xmath0 is given by the coupled homogeneous difference equations
@xmath5,\label{feq}\\ \hat f(p , q)&=&\frac{1}{6}\left [ f(p-1,q)+\hat f(p , q-1)+f(p+1,q)\right.\nonumber\\ & & \quad \left .
+ f(p+1,q+1 ) + \hat f(p , q+1)+ f(p-1,q+1)\right],\label{fhateq}\end{aligned}\ ] ] where @xmath6 denotes the field value at an even @xmath3 co - ordinate and @xmath7 denotes the field value at an odd @xmath3 co - ordinate .
this separation of even and odd field equations , which is not necessary for the problem on the square lattice , is central to our solution below .
now consider a source term at @xmath8 and absorbing boundaries at @xmath9 and @xmath10 . in section 3
we give the solution for the case where @xmath11 is odd .
the same methods can be used with slightly altered equations to obtain the solution for @xmath12 even .
the special case of the strip with @xmath13 has been detailed elsewhere @xcite . following mccrea and whipple @xcite we construct separate solutions , @xmath14 for @xmath15 and @xmath16 for @xmath17 .
the absorbing boundary conditions are thus given by @xmath18 the ommision of a subscript @xmath19 or @xmath20 in the above indicates that the same equations are satisfied by both @xmath21 and @xmath22 . at @xmath23
we have the matching conditions @xmath24 finally , with the inclusion of the source term , we have the coupled inhomogeneous field equations at @xmath23 , @xmath25,\\ 6\hat f_i(p , b)&=&6\delta_{p , a}(a \mbox{mod } 2)\nonumber\\ & & + \left[f_i(p-1,b)+\hat f_i(p , b-1)+f_i(p+1,b)\right.\nonumber\\ & & \left .
\quad + f_{ii}(p+1,b+1 ) + \hat f_{ii}(p , b+1)+f_{ii}(p-1,b+1)\right].\end{aligned}\ ] ] in section iii we present the solutions to eqs .
( 1)(12 ) .
the probabilities for the random walking particle to be absorbed at any specified point on one of the four boundaries , @xmath26 , are readily obtained by averaging over the nearest neighbour expectation values subject to the boundary conditions , eqs .
( 3)(8 ) .
for example , @xmath27;\nonumber\\ & & \qquad\quad k=1,\ldots \frac{m+1}{2}.\end{aligned}\ ] ]
we begin by solving the homogenous equations , eqs .
( 1),(2 ) , subject to the absorbing boundary conditions , eqs .
( 3),(4 ) .
the homogenous field equations separate , on using @xmath28 into @xmath29 where @xmath30 and @xmath31 are separation constants .
the two coupled equations for @xmath32 and @xmath33 can be readily decoupled into separate equations @xmath34 where @xmath35 is an integer .
this second order linear difference equation for @xmath36 on a lattice of even ( odd ) integers has the solution @xmath37 where @xmath38 and @xmath39 and @xmath40 are arbitrary .
it follows that the solution of @xmath41 ( with @xmath3 even ) corresponding to the boundary conditions , eqs .
( 3),(4 ) , is @xmath42 with @xmath43
the solution for @xmath44 ( with @xmath3 odd ) now follows from eq .
( 17 ) as @xmath45 equations ( 16 ) and ( 17 ) can also be decoupled for @xmath46 and @xmath47 resulting in the same fourth order linear difference equation in each case ; @xmath48 the zeroes of the corresponding characteristic quartic polynomial are in reciprocal pairs leading to the solutions @xmath49 where @xmath50 the coefficients @xmath51 may be chosen arbitrarily with the coefficients @xmath52 then determined by substituting eqs .
( [ qeq ] ) , ( [ qhateq ] ) into eq .
( 16 ) or eq .
( 17 ) . from eq .
( 17 ) we obtain @xmath53 where @xmath54 by combining eqs .
( 14),(15),(20),(22),(24),(25),(28),(29 ) we can write the solutions to the field equations , eqs .
( [ feq ] ) , ( [ fhateq ] ) , that satisfy the boundary conditions , eqs .
( 3),(4 ) , in the form @xmath55,\\ & & \nonumber\\ & & \hat f(p , q;a , b , c , d)=\sum_{j=1}^m\sin\left(\frac{\pi j p}{m+1}\right ) \cos\left(\frac{\pi j}{m+1}\right)\nonumber\\ & & \quad\times\left [ a_j{\mathrm{e}}^{\alpha q}({\mathrm{e}}^\alpha+1 ) + b_j{\mathrm{e}}^{-\alpha q}({\mathrm{e}}^{-\alpha } + 1)\right.\nonumber\\ & & \qquad\left .
+ c_j{\mathrm{e}}^{\beta q}({\mathrm{e}}^\beta+1)+ d_j{\mathrm{e}}^{-\beta q}({\mathrm{e}}^{-\beta}+1)\right],\end{aligned}\ ] ] where @xmath56 are arbitrary . before imposing the remaining absorbing boundary conditions , eqs .
( 5)(8 ) , we represent the solutions in the two regions : @xmath57 ; @xmath58 by @xmath59 the absorbing boundary conditions , eqs .
( 5)(8 ) , can be used to eliminate @xmath60 . after an appropriate re - normalization of the coefficients @xmath61 this yields @xmath62,\label{fi } \\ & & \nonumber\\ \hat f_i(p , q)&=&\sum_{j=1}^m\sin\left(\frac{\pi j p}{m+1}\right ) \cos\left(\frac{\pi j}{m+1}\right)\nonumber\\ & & \times\left[\hat a_i(q,\alpha,\beta)a_j+\hat a_i(q,-\alpha,\beta)b_j\right ] , \label{fihat}\\ & & \nonumber\\ f_{ii}(p , q)&=&\sum_{j=1}^m\sin\left(\frac{\pi j p}{m+1}\right)\nonumber\\ & & \times\left[a_{ii}(q,\alpha,\beta)\tilde a_j+a_{ii}(q,-\alpha,\beta ) \tilde b_j\right],\label{fii}\\ & & \nonumber\\ \hat f_{ii}(p , q)&=&\sum_{j=1}^m\sin\left(\frac{\pi j p}{m+1}\right ) \cos\left(\frac{\pi j}{m+1}\right)\nonumber\\ & & \times\left[\hat a_{ii}(q,\alpha,\beta)\tilde a_j+ \hat a_{ii}(q,-\alpha,\beta)\tilde b_j\right],\label{fiihat}\end{aligned}\ ] ] where @xmath63 and the function @xmath64 the coefficients defined in eq .
( 36 ) which are shown as implicit functions of @xmath65 and @xmath66 are also functions of @xmath67 via eqs .
( 21),(26),(27 ) . equations ( [ fi])([fiihat ] ) , are general solutions to the coupled homogeneous field equations , eqs .
( 1),(2 ) , that satisfy all of the absorbing boundary conditions , eqs .
( 3)(8 ) . by using the two matching conditions , eqs .
( 9),(10 ) , the four arbitrary constants , @xmath68 , can be reduced to two arbitrary constants , @xmath69 say .
the solutions in region ii can then be written as @xmath70,\\ & & \nonumber\\ \hat f_{ii}(p , q)&=&\sum_{j=1}^m\sin\left(\frac{\pi j p}{m+1}\right ) \cos\left(\frac{\pi j}{m+1}\right)\nonumber\\ & & \times\left[\hat a_{ii}^\star(q,\alpha,\beta)a_j + \hat a_{ii}^\star(q,-\alpha , \beta)b_j\right],\end{aligned}\ ] ] where @xmath71 and @xmath72 finally the remaining arbitrary constants @xmath69 are determined from the requirement that the solutions satisfy the coupled inhomogeneous field equations , eqs .
( 11),(12 ) .
this step is facilitated using the identity @xmath73 explicitly we find : + i ) @xmath74 even , @xmath75 ii ) @xmath74 odd , @xmath76 and @xmath77 our solution for the triangular lattice site expectation values with absorbing boundary conditions and a source point is finally given by eqs .
( 32),(33 ) in region i and eqs .
( 41),(42 ) in region ii .
the relevant quantities appearing in these equations are defined through the series of equations , eqs .
( 21),(26),(27),(36)(40),(43)(46),(48)(50 ) .
consider the case of a triangular lattice with @xmath78 and a source at @xmath79 ( fig .
2 ) . in the approximation to this problem using straight edge boundaries and the @xmath0 co - ordinate system of keberle and montet @xcite this problem corresponds to ; @xmath80 .
we have calculated expectation values at the nearest neighbour lattice sites around the source and absorption probabilities at the boundaries ; ( a ) using our exact results above , and ( b ) using the results of keberle and montet @xcite for the approximation to the problem their equations ( 7a ) , ( 7b ) , ( 7c ) . in this small model lattice system we found that the straight line boundary approximation provides reasonable results for expectation values near the source ( accurate to within a few percent ) but provides poor results for the absorption probabilities ( table 1 ) .
note that the ` absorption probabilities ' for the zig - zag boundary co - ordinates which are inside the straight line boundary of keberle and montet @xcite are not true probabilities in their solution .
@*5l @xmath81(a)&&&(b)&@xmath82&.012247&&@xmath83&.019609 + @xmath84&.035646&&@xmath85&.039949 + @xmath86&.054827&&@xmath87&.057904 + @xmath88&.063296&&@xmath89&.065834 + @xmath90&.056686&&@xmath91&.059209 + @xmath92&.039811&&@xmath93&.042817 + @xmath94&.020737&&@xmath95&.024583 + @xmath96&.005743&&@xmath97&.007903 + @xmath98&.026040&&@xmath99&.033624 + @xmath83&.013797&&@xmath84&.036214 + @xmath100&.069523&&@xmath101&.088151 + @xmath85&.021476&&@xmath88&.053888 + @xmath102&.005743&&@xmath103&.015276 + @xmath104&.040253&&@xmath105&.051612 + @xmath106&.015039&&@xmath107&.038562 + @xmath108&.059725&&@xmath109&.075927 +
although our solution is somewhat unwieldly , we have nevertheless solved the underlying field equations for random walks on a finite triangular lattice with a single interior source point and zig - zag absorbing boundaries and have thus calculated the associated absorption probabilities .
this problem was previously considered to be intractable @xcite .
we hope that our result will inspire further work in this area . | the problem of a random walk on a finite triangular lattice with a single interior source point and zig - zag absorbing boundaries is solved exactly .
this problem has been previously considered intractable .
# 1 epsf.sty |
auroral kilometric radiation ( akr ) bursts exhibit a wide variety of fine structure as seen on frequency - time spectra .
the cyclotron maser instability ( cmi ) @xcite is widely assumed to be the basic plasma mechanism responsible for the emission .
this mechanism originally assumed a loss - cone electron velocity distribution function , but in situ observations in the acceleration region have shown that a horseshoe or crescent distribution is more accurate @xcite .
the horseshoe distribution , which arises naturally for electron beams in the presence of inhomogeneous magnetic fields @xcite , provides a robust and efficient free energy source for the cmi mechanism , as shown both by model calculations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and in laboratory experiments ( e.g. , * ? ? ?
this mechanism has been applied not only to terrestrial akr emission , but to many astrophysical environments in which the requisite conditions ( low density , beamed electrons , inhomogeneous magnetic fields ) are thought to be present , e.g. planetary magnetospheres @xcite , stellar magnetospheres @xcite , and even relativistic jets in active galaxies @xcite .
detailed physical models for the rich variety of observed spectral signatures are still an active area of research .
akr fine structure exhibits many morphologies on time - frequency spectra , including slowing drifting or nearly stationary features lasting tens of seconds @xcite , drifting features which may be interacting with each other @xcite , and periodically modulated or banded emission @xcite .
although some akr bursts appear to be broadband emission extending over 100 s of kilohertz @xcite , the total power is dominated by highly time - variable fine structures , especially during periods of enhanced bursts @xcite , suggesting that perhaps all akr radiation is a superposition of narrowband , short duration fine structures .
@xcite first suggested that drifting akr fine structure may be due to localized sources rising and falling along auroral magnetic field lines , with emitted frequency equal to the local electron cyclotron frequency .
@xcite suggested that the emission is triggered by electrostatic waves which drift along the field at the local ion - acoustic speed . an early detailed model which attempts to explain akr fine structure is the tuned cavity model of @xcite in which the source region acts as a waveguide with sharp density boundaries .
the source emits radiation in normal modes analogous to an optical laser , resulting in narrowband emission which drifts as the wave packet propagates into regions of varying width .
a related idea is that of @xcite in which the cavity boundary is oscillating quasi - periodically . under these conditions , broadband radio photons
will be stochastically accelerated , resulting in quasi - monochromatic discrete tones , whose frequencies slowly drift with changes in the cavity geometry .
a difficulty with these models is that they require special conditions ( e.g. sharp density boundaries , oscillating walls ) which are not supported by observations @xcite .
several authors have reproduced akr fine structure using electromagnetic particle simulations to model the akr source region .
@xcite used a one - dimensional electromagnetic particle - in - cell ( pic ) code to model the cyclotron maser instability in a plasma with an inhomogeneous magnetic field .
they found that radiation is emitted in individual packets which combine to form drifting features that both rise and fall , consistent with some types of akr fine structure . both x- and
o - mode drifting fine structures are created , consistent with observations @xcite but not predicted in feedback models such as @xcite .
a weakness of their model is that predicted drift rates are generally much higher than observed drifts .
@xcite used a 2-dimensional pic code with model parameters based on in situ fast spacecraft measurements of the akr source region .
the simulated akr is most strongly amplified longitudinally and consists entirely of short - timescale fine structures .
the predicted bandwidth ( @xmath1 ) is somewhat larger than at least some observed akr fine structure .
more recently , pottelette and colleagues have investigated the possible connection between akr fine structure and electron holes @xcite and also tri - polar structures @xcite .
akr radiation from small packets ( elementary radiation sources ) associated with the holes are thought to be strongly amplified as the packets slow down and are reflected from the field - aligned electric field .
since the scale size of electron holes is very small ( a few debye lengths , @xcite ) , the observed narrow bandwidth of akr fine structures is easily accounted for .
however , since neither the observed speeds or direction of electron holes ( 1000 - 2500 km s@xmath2 always downward , @xcite ) are representative of fine structure frequency drifts , there is not a straightforward relationship between the observed properties of electron holes and the dynamics of the akr fine structure . in this paper
, we present evidence that a particular type of akr fine structure called striped or striated akr @xcite is triggered by ion solitary structures ( ion holes ) . in section 2 we present new observations of sakr bursts , including new bandwidth and angular beamwidth measurements . in section 3
we calculate the change in cmi gain for a density depleted region with a pre - existing horseshoe velocity distribution when perturbed by a passing ion hole .
we compare the observed properties of sakr with those calculated by radiation generated from the cmi instability in an ion hole . in section 4
we discuss some inferred properties of ion holes , such as lifetimes , probability of occurrence as a function of location , and radial dependence of speed , that can not be inferred from single spacecraft in situ measurements .
striated akr ( sakr ) consists of trains of narrowband drifting bursts with negative slopes in the range -2 to -20 khz s@xmath2 , corresponding to upward - traveling sources with speeds between 100 - 1,000 km@xmath3 .
figure [ fig - dynspec1 ] shows a typical dynamical spectrum of sakr bursts observed with the cluster wideband ( wbd ) plasma wave instrument @xcite on the four cluster spacecraft .
the bursts were observed in the 125 - 135 khz band between 16:14:20 - 16:14:35 ut on 31 august 2002 .
although the spectra are similar on all four spacecraft , there are clear differences in individual striations , indicative of angular beaming on the scale of the projected spacecraft separations , as shown in the inset at lower left .
also , individual bursts have very narrow bandwidths and are nearly always observed in groups with spacing between bursts of 30 - 300 ms . these characteristics are described in more detail below .
the occurrence probability of sakr emission is quite low at frequencies above 100 khz : we detect sakr bursts in less than 1% of all wbd spectra observed when the spacecraft was above 30magnetic latitude ( note that below this latitude , there is often shadowing by the earth s plasmasphere . ) the occurrence probabilities were computed by dividing the number of dynamic spectra ( length 52 sec ) for which sakr emission was clearly detected by the total number of spectra over a one year interval ( july 2002 to 2003 ) .
this overestimates the actual occurrence probability for sakr since we do not correct for the fractional time within each spectrum that sakr is present .
there is a strong inverse correlation with observing frequency as shown figure [ fig - histo - freq ] .
there are only few detections in several hundred hours of data in the 500 - 510 khz band ( detection probability @xmath4 ) , while @xmath5% in the 125 - 125 khz band .
@xcite also studied sakr emission using the plasma wave instrument ( pwi ) receiver on the polar spacecraft .
they also found an inverse correlation with observing frequency , with a 6% detection rate in the 90 khz band ( fig .
[ fig - histo - freq ] , shaded bar ) and comparable probabilities and overlapping frequencies . by contrast , non - striated akr emission is frequently detected , especially during sub - storm onset .
the occurrence probability of akr is nearly 30% for geomagnetic index k@xmath6 in the range 1@xmath7k@xmath8 3 , with the highest occurrence frequency in the winter polar regions @xcite .
sakr bursts have negatively sloped , nearly linear morphologies , at least within the normally sampled bandwidth ( 9.5 khz ) of the wbd receivers .
the majority of the slopes range between -2 khz s@xmath2 and -8 khz s@xmath2 . assuming the emission frequency is identified with the local electron gyrofrequency , the speed of the stimulator as a function of frequency and slope can be written @xmath9 where v is the stimulator speed ( km@xmath10 , _
@xmath11 _ is the observed slope of the sakr burst ( khz s@xmath12 , @xmath13 is the observed frequency ( khz ) , and we have assumed the source moves upward along a dipolar magnetic field line at magnetic latitude @xmath14 = 70 .
figure [ fig - histo - slope ] shows a histogram of the observed slopes of 650 sakr burst events observed in the 125 - 135 khz band along with the derived trigger speeds ( top x axis ) .
the mean slope is -5.6 khz s@xmath2 , with more than 90% of the slopes in the range -2 to -8 khz s@xmath2 .
the corresponding trigger speeds , using equation [ eq - speed ] , are in the range 76 303 km s@xmath2 , with a mean value 213 km s@xmath2 .
these results are similar to those reported by @xcite .
figure [ fig - dyn - spect2 ] shows a group of sakr bursts detected on one spacecraft while operating the wbd instrument in a wider bandwidth ( 77 khz ) mode . over this wider bandwidth ,
the striations are not linear , but rather have a curved , frequency - dependent slope .
the curve can be fit by assuming a source radiating at the local electron cyclotron frequency and moving with constant speed along a field line in a dipolar magnetic field .
the overlaid white line shows the expected trace for a source moving upward at a constant velocity of 300 km@xmath3 .
sakr bursts with similar frequency - dependent behavior are seen in plate 1 of @xcite .
the altitude range of the sakr locations are shown on the right y - axis .
for this example , individual bursts originated near 6500 km altitude and moved upward at 300 km@xmath3 to an altitude near 8100 km , implying a lifetime of several seconds .
we have examined several other sakr events with wide bandwidth and have found that all exhibit similar lifetimes and have nearly constant speed , although the speed varies with epoch .
sakr bursts are almost always detected superposed on broader band akr emission .
this can be clearly seen in both figure [ fig - dynspec1 ] and [ fig - dyn - spect2 ] , as well as in plate 1 of @xcite .
this appears to be a universal characteristic of sakr , except in cases where the sakr intensity is so low that underlying broadband emission may have been undetected .
background akr emission coeval with sakr bursts is consistent with the hypothesis that ion holes trigger the sakr bursts ( section 3.2 ) since the cmi gain in the source region is significant even in the absence of the ion holes .
one of the most unusual features of sakr bursts is their extraordinarily narrow bandwidth .
the top panel of figure [ fig - bw ] shows a single isolated sakr burst .
the middle panel shows the same feature , but after de - trending by subtraction of a ` chirp ' signal of best - fit constant negative slope .
the lower panel shows the full width at half maximum of a gaussian fit along the frequency axis of the de - trended signal after summing in time intervals of 37 ms each .
the resulting bandwidths , in the range 15 22 hz , are much narrower than previously reported either observationally ( e.g. @xcite ) or resulting from model calculation of akr emission ( e.g. @xcite ) although @xcite also reported akr bandwidths as small as 5 hz .
( in the latter paper , the bursts do not appear to be sakr emission ) .
other sakr bursts we have examined have bandwidths ranging from 15 to 40 hz .
the narrow bandwidth implies a small source extent along the z ( b - field ) direction , viz , @xmath15^{\frac{4}{3}}\left [ { \frac{\delta f}{10~hz } } \right]\ ] ] where @xmath16z is the source extent along the b field and @xmath16f is the observed bandwidth . for @xmath17 khz and @xmath16f = 20
hz we obtain @xmath16z = 0.76 km .
this is considerably smaller than the lateral extent of the akr source region and indicates the trigger for sakr must have a dimension along the magnetic field of order 1 km .
the cluster spacecraft array has the unique ability to simultaneously sample the flux density of individual akr bursts at four widely separated points in space . by comparing the flux density on pairs of spacecraft
, we can estimate the average angular beam size of individual bursts .
the beaming probability is defined using the following algorithm .
for each pair of spacecraft during a given observation , we calculate the projected angular separation between spacecraft as seen from a location situated above the magnetic pole of the hemisphere being observed , at a height corresponding to the electron cyclotron frequency at the center frequency of each observing band ( e.g. , 2.35 r@xmath18 for the 125 - 135 khz band ) .
this constitutes an average akr location in that band without regard to location on the auroral oval .
we next correct for differential propagation delay by shifting the waveform data from each spacecraft to the distance of the nearest spacecraft .
we then divide the time - frequency spectrum from each spacecraft pair ( 52 sec duration for 125 - 135 khz band ) into data ` cells ' 19 msec x 53 hz in size .
this window was chosen to match the observed bandwidth of sakr ( cf .
section 2.4 ) but the beaming results are not very sensitive to the data window size .
we computed the angular beam size with data windows factors of two smaller and larger , and the results did not differ significantly .
the intensities are normalized to correct for differing distances between individual spacecraft and the akr source .
we then omit from further analysis all data cells whose intensities are below a threshold , arbitrarily chosen to be 10 db below the maximum flux density for that spectrum . finally , we compare intensities in each pair of data windows , assigning a weight 1 to pairs for which the intensities are within 10 db of each other , and 0 otherwise .
the overall beaming probability for each angular separation interval is the sum of the beaming weights divided by all cell pairs .
note that this scheme is susceptible to overestimation of the beaming probability , since it is possible that independent akr sources will illuminate separate spacecraft at the same frequency and time interval .
this ` confusion ' problem is smaller for sakr emission since it is often clear from the burst morphology on a time - frequency spectrum that only one source contributes to a given data cell at one time , whereas with normal akr there are very often several intersecting sources which contribute to a given data cell . in figure [ fig - angbeam ]
we plot the beaming probability versus angular separation for 651 sakr bursts in the 125 - 135 khz and 250 - 260 khz bands .
we have fitted a gaussian function to the 125 khz band observations using a least - squares fitting algorithm .
the resulting full width at half maximum ( fwhm ) angular size is _
@xmath19 _ = 5.0 ( solid angle @xmath20 = 0.006 sr ) .
this is surprisingly small compared with most previously published observations of akr beam size ( e.g. @xcite ) , who reported beaming solid angles of 4.6 sr and 3.3 sr at 178 and 100 khz respectively .
they made angular beaming estimates by comparing time - averaged spectra observed using two satellites ( hawkeye and imp-6 ) which simultaneously observed akr bursts while the spacecraft were both over the same polar region .
since the time resolution used for the hawkeye - imp6 spectrum comparison ( several minutes ) far exceeds the time - scale of individual akr bursts , their measured angular beam is actually a measure of the ensemble - averaged sky distribution of akr bursts over a several minute time - scale rather than the angular beam size of individual akr emission sources .
this is the confusion problem mentioned above .
an important unanswered question is the 2-dimensional structure of the sakr burst angular beam pattern : it is asymmetric , or perhaps a hollow cone as suggested by @xcite ?
since the spatial frequency coverage of the cluster spacecraft array is often nearly one - dimensional at high magnetic latitude , it is difficult to analyze the 2-dimensional structure of individual bursts .
the angular beaming probability plot shown in figure [ fig - angbeam ] includes measurements over a range of baseline orientations .
however , a preliminary analysis of beaming probability grouped by baseline orientation did not reveal any obvious trends .
we are presently analyzing a much a larger dataset consisting of a large variety of akr emission and wider range of baseline orientations to investigate the 2-dimensional structure of the akr angular beam .
the angular beaming observations , combined with measured flux density , allow a direct estimate of the average intrinsic power of individual sakr bursts . for intense sakr bursts ,
the observed square of the electric field intensity is @xmath21 at a source - spacecraft distance @xmath22 = 10r@xmath18 . converting to flux density , we obtain @xmath23 where @xmath24 = 377 ohms is the impedance of free space .
this flux density range is about one hundred times smaller than the flux density of intense akr bursts reported by @xcite .
this is consistent with the conjecture that intense akr bursts are the sum of many spatially distinct elementary radiation sources @xcite the power emitted at the akr source is the isotropic power corrected by the angular beamsize of an individual sakr burst @xmath25 where _
@xmath26 _ @xmath27 20 - 50 hz is the bandwidth of a burst , and _
@xmath20 _ @xmath27 0.006 sr , is the solid angle of the emission beam .
using these values , the resulting power is p @xmath27 1 10 w , much smaller than the previous estimates of p @xmath27 10@xmath28 - 10@xmath29 w for a single elementary radiator @xcite .
ion solitary structures , also known as ion holes , are small - scale ( @xmath301 km ) regions of negative electrostatic potential associated with upgoing ion beams .
they are seen in spacecraft electric field measurements as symmetric bipolar parallel electric field structures with amplitudes 10 500 mv m@xmath2 and timescales 3 - 10 ms .
they were first detected in s3 - 3 spacecraft observations @xcite and have been subsequently been studied using in situ measurements in the acceleration region by several authors ( e.g. @xcite ) .
ion holes travel upward at speeds between 75 - 300 km@xmath3 @xcite ( although @xcite argue for somewhat higher speeds , in the range 550 - 1100 km@xmath3 ) . the width of the waves increase with amplitude @xcite which is inconsistent with small amplitude 1-d soliton models , but which supports a bgk - type generation mode @xcite .
akr radiation arises from wave growth resulting from the interaction of a radiation field with an electron velocity distribution having a positive slope in the direction perpendicular to the magnetic field ( @xmath31/@xmath32v@xmath33 0 ) .
the condition for waves of angular frequency @xmath34 and wave normal angle @xmath19 to resonate with electrons with angular frequency @xmath35 and velocity @xmath36 is given by @xcite @xmath37 where @xmath38 is the electron cyclotron frequency , @xmath39 is an integer , and @xmath40 is the lorentz factor of the relativistic electrons .
the cyclotron maser instability ( cmi ) is the case @xmath41=1 .
if we assume that the electrons are mildly relativistic , as is observed in the akr source region ( e@xmath42 10 kev , so @xmath431.01 ) , we can expand the lorentz factor to obtain the equation for a circle in velocity space @xmath44 where the resonant circle s center is on the horizontal axis displaced by @xmath45 and the radius of the resonant circle is @xmath46^{\frac{1}{2}}\ ] ] analysis of recent fast observations of akr emission in the source region @xcite provides evidence , based on wave polarization , that the akr k - vector direction at the source is nearly perpendicular to the b field ( @xmath47 ) .
the e - field of the wave is polarized perpendicular to the ambient magnetic field which indicates that the wave is purely x - mode @xcite . in the following ,
we explicitly assume @xmath48 , so that the radius of the resonance circle becomes @xmath49 where @xmath50 @xmath51 .
the growth rate of the cmi mechanism is given by calculating the imaginary part of the angular frequency , @xmath52 where the @xmath53 integral is performed on the closed circular path given by equation 4 .
the electron velocity distribution in the upward current acceleration region has been measured _ in situ _ by both viking @xcite and fast @xcite spacecraft .
it consists of an incomplete shell or ` horseshoe ' shape in velocity space .
the density of cold electrons ( e @xmath54 1 kev ) in this region is much smaller than the hot electron population which comprises the horseshoe component @xcite . in this paper
we assume that all electrons are in the hot component .
we have modeled the observed velocity distribution using a simple analytic functional form @xmath55\ ] ] where _
@xmath16 _ is the horseshoe width , @xmath56 the horseshoe radius and the loss - cone function @xmath57 @xmath58 where @xmath59 is a dimensionless scaling factor and @xmath60 is the characteristic opening angle of the loss cone .
the model velocity distribution function is shown in figure [ fig - dist-2x1 ] along with a measured velocity distribution function from fast @xcite . , and inside an ion hole ( @xmath61 .
black circle is the cmi resonant circle .
the @xmath62direction is the horizontal axis , earthward to left .
bottom panel : weighted partial derivative @xmath31/@xmath32v@xmath63 outside an ion hole ( @xmath64 , and inside an ion hole ( @xmath65 .
the arrows indicate the parts of the growth line integral ( equation 9 ) enhanced inside the hole . ] ) .
the inset at lower right shows a simple spherical model of an ion hole with inward electric field .
( b ) calculated parallel velocity vs. distance from an ion hole for an electron passing through an ion hole with an impact parameter 1 km and initial velocities 20,000 km s@xmath2 ( dotted line ) and 30,000 km@xmath3 ( solid line ) .
the assumed hole speed is 300 km@xmath3 . ]
the range of possible interactions between background electrons and ion holes is very complex and varied @xcite .
these include trapping of electrons between ion holes , excitation and trapping of high - frequency langmuir waves from electron streams resulting from ion hole collisions , and modification of the ion holes via the ponderomotive force . in this paper
we will tacitly assume that the ion holes are well - spaced and non - interacting , and that the only significant effect on the electron population is a transient speed decrease as the electrons traverse the hole s negative potential well .
this assumption is motivated by the relatively simple structure of sakr bursts and by the success in using this assumption to explain all significant observed properties of sakr bursts .
other more complicated forms of akr fine structure seen on dynamical spectra may well involve one or more of the complex interactions mentioned above .
since the electrons are magnetized ( gyro - radius r@xmath6650 m , much smaller than the ion hole parallel scale size r@xmath671 km ) , the ion s hole s effect on the perpendicular velocity component is negligible ( a small * * e**x*b * drift , v@xmath68 40 m s@xmath12 .
however , the parallel velocity component will experience a significant decrease as the electrons are repelled by the negative potential well of the hole . to determine the magnitude of the effect
, we have modeled the observed electric field structure of an ion hole using the analytic form @xmath69 figure [ fig - hole - efield]a shows an observed e - field of an ion hole along with a plot of the model e - field with @xmath70 = 500 mv m@xmath2 , and @xmath71 ms .
we used this model to calculate the velocities of electrons as they traverse the hole for a variety of impact parameters and initial speeds .
for initial parallel speeds of 20,000 and 30,000 km@xmath3 , the speed of an electron traversing an ion hole at a minimum distance of 1 km from the center is shown in figure [ fig - hole - efield]b . as expected , the parallel velocity briefly decreases , so that the horseshoe velocity distribution is ` squeezed ' on the parallel ( horizontal ) axis , as shown in figure [ fig - elec - dist-4by1]b .
the speed decrease is proportional to @xmath72 as expected by a simple energy conservation argument : for an electric field e and ion hole diameter l@xmath73 , the potential well of the hole is approximately @xmath74e@xmath75l@xmath73 .
the electron loses kinetic energy as it traverses the potential well , so that energy conservation requires @xmath76 solving for the velocity change , @xmath77 for an ion hole electric field e @xmath27 300 mv m@xmath2 , diameter l@xmath78 @xmath27 2 km , and @xmath56 = 20,000 km@xmath3 , the expected speed decrease is @xmath79 5,000 km@xmath3 , in good agreement with the exact calculation . in order to calculate the power gain in the perturbed region , we need an estimate of the convective growth length @xmath80 where @xmath81 is the group velocity of the wave and @xmath82 is the growth rate .
the group velocity is very sensitive to the ratio of plasma to gyro - frequency and the detailed velocity distribution function . as a first approximation ,
we have used the cold plasma dispersion relations to estimate the group velocity @xmath83 for frequencies near the r - mode cut - off frequency .
we find that @xmath84 is of order 300 - 1000 km s@xmath2 , similar to the estimate used by @xcite . for a maximum growth rate @xmath85 5000 s@xmath2 ( at @xmath86 = @xmath87 khz ) ,
the corresponding convective growth length is l@xmath88200 m .
hence , there are approximately 5 - 10 e - folding lengths in the region of the iss , assuming a perpendicular physical scale of order 1 - 2 km .
this results in a maximum power gain of e@xmath89 43 - 87 db .
the power gain required to amplify the background radiation to observed levels of normal ( wideband ) akr was estimated by @xcite to be e@xmath90 , which is at the high end of our calculation . however , as discussed above , since sakr emission is highly beamed , the required power is a factor of @xmath27100 smaller , so the requisite power gain is closer to e@xmath91= 65 db .
this is comfortably within the range predicted by an ion hole trigger .
figure [ fig - cmi - growth ] illustrates the growth rate and power gain both inside and exterior to an ion hole for a range of relativistic electron energies and plasma densities .
figure [ fig - cmi - growth]@xmath92 show the growth rate and power gain respectively for a horseshoe velocity distribution with a radius @xmath93 ( e = 5.7 kev ) and @xmath94 = 0.1 , 0.2 and 0.5 @xmath95 , while figure [ fig - cmi - growth]@xmath96 shows the growth rate and gain for @xmath94 = 0.2 @xmath95 and a range of radii @xmath97 = 0.10 , 0.15 , and 0.20 ( e = 2.5 , 5.7 , 10.2 kev ) .
these values are typical of the values of electron density and energy observed in the auroral density cavity @xcite .
there are several noteworthy features of these plots : 1 .
the range of electron densities which result in substantial cmi gain is very limited .
the maximum density is constrained by the condition that the r - mode cutoff frequency is less than the electron cyclotron frequency .
this can by expressed by the inequality @xmath98 ^{1/2}\ ] ] where @xmath99 and @xmath100 are the electron plasma and cyclotron frequencies respectively , and @xmath101 is the lorentz factor .
this inequality can also be written @xmath102 where @xmath103 is the rest - mass energy of the electron ( 511 kev ) .
for example ,
for @xmath104 = 5 kev and @xmath13= 125 khz , we find @xmath105 .
on the other hand , for very low densities ( @xmath106 ) , the cmi gain decreases dramatically ( cf . fig .
[ fig - cmi - growth]@xmath107 ) 2 .
the gain is a sharply peaked function of frequency .
for example , for @xmath94 = 0.3 @xmath95 , @xmath108 5.7 kev ( fig .
[ fig - cmi - growth]@xmath109 ) , a dynamic range of 30 db near the peak ( approximately that observed on dynamic spectra of sakr bursts ) corresponds to a fractional bandwidth @xmath110 0.0005 , or @xmath111 50 hz at @xmath13 = 125 khz , in good agreement with observed bandwidths ( section 2.4 ) .
these extremely narrow bandwidths are characteristic of cmi gain curves with partial - ring velocity distributions @xcite .
3 . even outside the ion hole
there is substantial cmi gain under some favorable conditions ( e.g. 80 db for the middle plot of figure [ fig - cmi - growth]@xmath65
. this may be the explanation for the background akr seen in figure [ fig - dyn - spect2 ] and discussed in section 2.3 . given the rather small volume of an ion hole , it is reasonable to ask if there is sufficient free energy available from the resonant electrons to power the observed sakr emission .
as shown in section 2.5 , the observed flux levels and angular beamwidths of sakr emission correspond to radiated powers @xmath112 w. using a an ion hole scale size @xmath113 1 km and speed @xmath114 , and mean electron density and energy @xmath115 , @xmath116 = 5 kev respectively , the ratio of radiated power to the electron kinetic energy traversing a hole per unit time is @xmath117 this range of conversion efficiencies is similar to previous models of akr emission ( e.g. , * ? ? ? * ) and indicates that while a single ion hole extracts relatively little energy from the ambient electron population , a train of hundreds of holes traversing the same region could account for a significant modification of the electron distribution function .
if sakr bursts are stimulated by upward traveling ion holes in the acceleration region of the magnetosphere , then observations of these bursts provides a new technique to study characteristics of ion holes such as lifetimes and relative number and speed as a function of altitude , that are impossible to measure with single _ in situ _ spacecraft . 1 .
the 77 khz bandwidth observations of sakr bursts indicate that ion holes propagate upward for more than 1,000 km , implying lifetimes of a few seconds .
this is much longer than estimates of solitary wave lifetimes derived from pic simulations of solitary waves generated by the two - stream instability @xcite , which have lifetimes @xmath118 5 - 75 ms .
the majority of ion hole speeds derived from sakr observations are in the range 75 - 400 km s@xmath2 ( fig .
[ fig - histo - slope ] ) .
these speeds are in very good agreement with _ in situ _ measurements of ion hole speeds from polar ( 75 - 300 km s@xmath2 ) at altitudes between 5500 km - 7000 km @xcite .
however , @xcite finds somewhat higher ion hole speeds ( 550 - 1100 km s@xmath2 ) based on data from the fast satellite at altitudes between 3000 km - 4000 km .
it is possible that at speeds above @xmath119600 km s@xmath2 ( slopes @xmath12020 khz s@xmath2 ) sakr bursts , especially in closely spaced groups , would be undetected since they would merge into quasi - continuous emission on dynamic spectra .
3 . fits to sakr bursts assuming a constant speed source , ( fig .
[ fig - dyn - spect2 ] ) indicate that ion holes propagate at nearly constant speed for their entire lifetime .
some numerical simulations @xcite show a significant change in ion hole speed as they evolve .
this is not supported by our observations .
sakr bursts , and hence ion holes , are much more common at higher altitude , being more than one hundred times as common at 10,000 km ( @xmath121 90 khz ) than at 3,200 km ( @xmath121 500 khz ) altitude , assuming ambient conditions favorable to the generation of sakr are not dissimilar in this altitude range .
since sakr bursts are almost always detected in groups with typical spacing 30 - 300 ms . , ( section 2.6 ) , this likely also applies to ion holes .
both laboratory plasma experiments @xcite and observations of ion holes in the magnetosphere ( e.g. , * ? ? ?
* ) show trains of ion holes with spacings similar to the sakr bursts .
finally , the uniformity of sakr intensity and bandwidth over a large frequency range ( fig.[fig - dyn - spect2 ] ) implies that there is little evolution of the electric field intensity or spatial structure of ion holes over their lifetime .
this paper summarizes the observed properties of a distinct form of auroral kilometric radiation fine structure called striated akr , first described by @xcite .
we present new observational results using the wbd instrument on cluster which characterize the bandwidth and angular beamsize of individual sakr sources , as well as derived properties such as intrinsic power and speed along the magnetic field . assuming the frequency of sakr bursts
can be identified with the local electron cyclotron frequency , the speed and direction of sakr sources calculated from their observed drift rates are very similar to those observed for ion solitary structures ( ion holes ) in the upward current region of the magnetosphere . hence , we investigated whether sakr bursts could be the result of enhancement of the cyclotron maser instability by ion holes .
using observed electric field signatures of ion holes in this region , we calculate the perturbation caused by the passage of an ion hole on a horseshoe electron velocity distribution in a dilute ( @xmath122 @xmath95 ) plasma .
the cyclotron maser instability is strongly enhanced inside the ion hole , with power gain exceeding 100 db in a narrow frequency range just above the x - mode cutoff frequency .
these characteristics are in excellent agreement with the observed bandwidth , speed , direction , and flux density of sakr bursts .
alternative suggestions involving other types of solitary structures to explain akr fine structure , such as electron holes @xcite or tri - polar structures @xcite , are not consistent with sakr properties , although they may be important in other types of akr fine structures .
if sakr bursts are in fact triggered by ion holes , a number of derived properties of ion holes can be deduced which would be difficult or impossible to obtain using _ in situ _ measurements .
these include average lifetimes ( a few seconds ) , evolution of propagation speed ( nearly constant over lifetime of hole ) , and relative numbers versus location ( much more common high in the acceleration region than near the base ) .
gurnett , d. and r. anderson , the kilometric radio emission spectrum : relationship to auroral acceleration processes , _ physics of auroral arc formation _ , geophysical monograph series , vol . 25 .
akasofu and j.r . kan .
washington dc : american geophysical union , 341 - 350 , 1981 .
pritchett p. l. , r. strangeway , r. ergun , and c. carlson . , generation and propagation of cyclotron maser emissions in the finite auroral kilometric radiation source cavity , _ j. geophys .
res_. , vol .
107 , a12 , doi:10.1029/2002ja009403 , 2002 . | we describe the statistical properties of narrowband drifting auroral kilometric radiation ( striated akr ) based on observations from the cluster wideband receiver during 2002 - 2005 .
we show that the observed characteristics , including frequency drift rate and direction , narrow bandwidth , observed intensity , and beaming angular sizes are all consistent with triggering by upward traveling ion solitary structures ( ` ion holes ' ) .
we calculate the expected perturbation of a horseshoe electron distribution function by an ion hole by integrating the resonance condition for a cyclotron maser instability ( cmi ) using the perturbed velocity distribution .
we find that the cmi growth rate can be strongly enhanced as the horseshoe velocity distribution contracts inside the passing ion hole , resulting in a power gain increase greater than 100 db .
the gain curve is sharply peaked just above the r - mode cut - off frequency , with an effective bandwidth @xmath050 hz , consistent with the observed bandwidth of striated akr emission .
ion holes are observed _ in situ _ in the acceleration region moving upward with spatial scales and speeds consistent with the observed bandwidth and slopes of sakr bursts .
hence , we suggest that sakr bursts are a remote sensor of ion holes and can be used to determine the frequency of occurrence , locations in the acceleration region , and lifetimes of these structures . |
recently various scenarios related to in - medium modifications of the properties of the @xmath3-meson have been proposed @xcite to explain the apparent increase in the production of dilepton pairs with an invariant mass around the @xmath3 meson in heavy ion collisions @xcite . to really be able to extract these signals , possibly due to a decrease in the @xmath3 mass as a result of chiral symmetry restoration
, the description of the fundamental processes has to be on a firm ground and its inherent approximations have to be well understood . in this paper
we will investigate the elementary process of dilepton pair production off nucleon - nucleon collisions .
two ingredients are of importance here .
first there is the question of the nucleon - nucleon form factor .
vector meson dominance suggests a form which naturally gives a significant increase in the cross - section around the @xmath3 mass .
however , while the concept is well established for the pion formfactor , for the nucleon formfactor its validity is much less clear @xcite . in this paper
we will focus on a second point , the off - shell character of the @xmath2 interaction . from real photon bremsstrahlung
one knows that already at pion threshold off - shell effects are crucial to describe the data . for virtual photon
bremsstrahlung one might expect the effects to be even larger since the typical energies involved are much higher than for real photon bremsstrahlung and one thus can reach kinematics which are further off shell . in this paper
we will present a calculation based on a @xmath0 matrix which includes @xmath1 degrees of freedom and fits the @xmath2 data up to 2 gev .
we will compare our results with a previous calculation that uses a one boson exchange fit to the @xmath2 on - shell data to find its off - shell behaviour @xcite .
the quantity of interest is the differential cross - section for production of lepton pairs with an invariant mass @xmath4 .
it is most easily calculated in the c.m .
frame of the incoming nucleon pair @xmath5^\frac{1}{2}.\end{aligned}\ ] ] in this expression @xmath6 is the elementary charge , @xmath7 the nucleon and electron masses respectively , @xmath8 the total incoming invariant mass and @xmath9 the momentum of the incoming nuclei . also , @xmath10 is the momentum of the outgoing nucleons relative to each other , its value is determined by energy conservation .
the final nucleon pair thus has momenta @xmath11 , the energies of the final nucleon pair are denoted by @xmath12 . in the integration over the solid angle of @xmath10 , @xmath13 is taken relative to the photon momentum , @xmath14 .
a similar notation is used for the dilepton pair , their momenta are @xmath15 , with corresponding energies @xmath16 .
furthermore , @xmath17 is the zero - th component of the photon momentum , @xmath18 .
@xmath19 is the leptonic tensor describing the decay of the virtual photon into the dilepton pair , @xmath20.\ ] ] @xmath21 is the matrix element for producing a virtual photon with rest mass @xmath4 off a nucleon - nucleon collision .
the diagrams we include in the calculation of this amplitude are shown in fig .
[ diagrams ] .
we have the nucleon single - scattering ( [ diagrams]a ) , rescattering ( [ diagrams]b ) and @xmath1 single - scattering ( [ diagrams]c ) and rescatter diagrams ( [ diagrams]d ) .
we include the same diagrams as in a previous study of real - photon bremsstrahlung @xcite , the expressions given there are readily generalized to the case of virtual photon production . for the @xmath22 vertex we again use the standard form without formfactors @xmath23 where we defined the photon momentum @xmath24 to be outgoing .
the form of the @xmath25 vertices is less well established .
following jones and scadron @xcite we have the general form @xmath26 in these expressions the index @xmath27 is to be contracted with an index of the @xmath1 propagator , @xmath28 is the total momentum and @xmath29 is third component of the isospin transition matrix for coupling an isospin 3/2 to an isospin 1/2 particle .
the @xmath30 component does not contribute in the case of a real photon ( @xmath31 ) and thus experimental input on this quantity is very limited .
one attempt is the calculation of nozawa and lee @xcite of electroproduction of pions off the nucleon .
however , using the values nozawa and lee employ we found that results were insensitive to the contribution from this vertex part and we can safely ignore this contribution .
the most elaborate determinations of the @xmath32 and @xmath33 coupling constants come from fitting the @xmath34 and @xmath35 multipole data on the photoproduction of pions off nucleons @xcite . in judging these numbers we have to realize that in our calculation we treat the nucleon background differently .
it is therefore the most realistic to take values from calculations where the nucleon background contribution in the @xmath36 channel is implicitly included in the coupling constants . in such an approach jones and scadron find @xmath37 ( gev@xmath38 ) and @xmath39 ( gev@xmath40 ) , which we will use in our calculation .
this is very close to the values found in a recent analysis of lee @xcite : @xmath41 ( gev@xmath38 ) and @xmath42 ( gev@xmath40 ) . with these numbers we have to bear in mind that the dominant contribution of the vertex comes from the @xmath43 part and
we can order the sets of values according to the value of @xmath32 , the accompanying value of @xmath33 has not much influence on the results .
an independent method to find the values is assuming vector - dominance .
then the values are determined by the ratio @xmath44 . using the values of ter haar @xcite we find @xmath45 ( gev@xmath38 ) and @xmath46 ( gev@xmath40 ) .
this value is on the low end of the range of values found in pion - photoproduction . in ref .
@xcite a value taken from the decay width of the @xmath1 into a photon and nucleon is used : @xmath47 ( gev@xmath38 ) and @xmath46 ( gev@xmath40 ) . in calculating
the diagrams we use a @xmath0-matrix that is based on the paris potential and additionally contain @xmath1-degrees of freedom .
including @xmath1 degrees of freedom is essential for describing the inelastic channels and resonant structures in the phase - shifts .
the resulting @xmath0-matrix describes the nucleon - nucleon data well up to 2 gev @xcite .
other calculations of dilepton production use a less sophisticated description of the @xmath2 interaction .
one approach is to develop a soft photon approximation for virtual photon bremsstrahlung as is done in ref .
another possibility is to parametrize the on - shell nucleon scattering data by means of an one boson exchange interaction , and use this parametrization to extrapolate to the off - shell matrix elements which enter in the calculation of the matrix element @xmath21 @xcite .
this method has certain limitations , first of all , the nn - scattering amplitude is not unitary due to its obe form . also , in the fit to the @xmath2 data in the approach of ref .
@xcite the on - shell equivalence of the pseudo - scalar and pseudo - vector coupling makes the parameters independent of whether one chooses a pseudo - scalar or pseudo - vector pion - nucleon - nucleon vertex .
however , the results for dilepton production do depend on this choice .
although the contribution from the electric part of the vertex remains the same @xcite , the contribution from the magnetic part of the vertex is affected by this choice . using a @xmath0-matrix one
has a more realistic description of the off - shell character of the @xmath2 interaction , since the off - shell amplitudes enter in the t - matrix equation and these are thus implicitly constrained by the fit to the @xmath2 data .
for example , the degeneracy of the pseudo - scalar / pseudo - vector choice for the pion coupling is resolved to a large extent : one has to use different parameters to obtain the same fit .
using a sophisticated description of the @xmath2 interaction has ramifications on other parts of the model .
the current arising from the nucleonic diagrams ( figs .
[ diagrams]a , b ) is not gauge - invariant since we do not include negative energy states in the intermediate propagators of the @xmath0-matrix .
note that the @xmath1-decay diagrams are gauge - invariant due to the gauge invariant @xmath25 vertex . for real photon bremsstrahlung
this is not too big a problem . in that case
the current is gauge invariant up to order @xmath48 in the photon momentum .
moreover , an actual calculation of the negative energy state contributions shows them to be small @xcite . due to the fact that the matrix element for virtual photon bremsstrahlung depends on two independent quantities
( the mass of the virtual photon and its momentum ) the situation for virtual photon - bremsstrahlung is less favourable . only in the c.m .
frame of the virtual photon one can show that the matrix element is model independent only up to order @xmath49 .
one also finds rather large differences between the various low energy theorems @xcite .
this all implies that the issue of gauge invariance is more important for dilepton pair production than it is for real photon bremsstrahlung .
the problem of a non - gauge invariant current is also encountered in models of @xmath50 reactions .
various schemes have been devised to render a calculated current gauge invariant @xcite .
of course neither scheme is unique and will lead to different results which can be related to the choice of a particular gauge @xcite .
having noticed that our @xmath51 is particulary large as compared to the spatial components we decided to use the method of refs .
@xcite and express @xmath51 in terms of @xmath52 to obtain a conserved current : @xmath53 the resulting amplitude then trivially fullfils the gauge - invariance condition .
this choice has the advantage that in this scheme the results for real - photon bremsstrahlung are unaffected since this process is independent of @xmath51 .
with the model described before we calculated the differential cross - section for production of dilepton pairs with invariant mass m for proton - proton collisions at 1 gev laboratory energy .
we restrict ourselves to proton - proton collisions since in neutron - proton collisions meson - exchange - currents usually dominate , but are very elaborate to calculate in a @xmath0-matrix approach .
we think that studying the pp reaction provides a clearer picture of the additional dynamics introduced by the use of a @xmath0-matrix .
we choose a laboratory energy of 1 gev to still be in the range where the @xmath0-matrix provides a detailed description of the phase - shifts . at higher energies the total and inelastic cross - sections
are still very well reproduced , but the reproduction of more sensitive observables like the polarization cross - sections is less satisfactory @xcite . since the latter is also true for other models that fit the @xmath2 data to high energies using a @xmath0-matrix formalism @xcite , one might argue that the description of @xmath2 scattering in terms of nucleon and meson degrees of freedom in terms of a @xmath0-matrix starts to break down and one has to include additional physics , like explicit quark degrees of freedom .
the results are displayed in fig .
[ results_1 ] . comparing our nucleonic contribution ( dash - dot - dot ) with the one of schfer _ et al . _
( dash - dot ) we see a rather large difference . at low @xmath4
our cross - section is around a factor 2 larger than the one of schfer . at higher
@xmath4 the situation is reversed . there
the cross - section of schfer is larger .
this can be largely attributed to the choice of a pseudo - scalar pion vertex .
taking a pseudo - vector pion coupling the result of schfer is reduced by a factor 10 and is again smaller than our result .
the soft - photon result of ref .
@xcite ( short - dashed line ) is even lower than the schfer result , and we can conclude that in our calculation the additional off - shell information we include leads to an enhancement of the nucleonic part of the cross - section . at the energies under consideration
the total cross - section is dominated by the @xmath1 contributions , even more than we found for the real - photon bremsstrahlung case @xcite . to show the dependence of the total results on the strength of the @xmath54 vertex we performed calculations with a strong coupling @xmath37 ( gev@xmath38 ) and @xmath39 ( gev@xmath40 , dotted line ) and with the relatively weak vector dominance value @xmath45 ( gev@xmath38 ) and @xmath46 ( gev@xmath40 , dashed line ) .
the result of schfer @xmath47 ( gev@xmath38 ) and @xmath46 ( gev@xmath40 ) is represented by the solid line .
the result with the weak coupling is about 25 % smaller than the one with the strong coupling . as stated before , these values are on the high and low end of the spectrum found in various pion - photoproduction models and thus represent a natural measure of the theoretical uncertainty .
the results of schfer _ et al .
_ nicely fall in between and the overall agreement between both models is reassuring .
it remains to be seen whether the similarities hold when looking at more refined observables , like e.g. anisotropies of the dilepton pairs @xcite .
another point of additional investigation is the influence of form - factors
. a straightforward application of vector dominance will give us simply an upward shift in the cross - section , increasing with increasing invariant mass @xmath4 .
however , a more detailed calculation of the form - factors , as performed in ref .
@xcite shows that the vector dominance assumption is too general for the nucleonic form - factor . in conclusion
, we presented a calculation of dilepton production at 1 gev laboratory energy using a @xmath0 matrix that includes @xmath1 degrees of freedom and provides an excellent fit to the @xmath2 scattering data .
we argued that the use of a @xmath0 matrix gives a more reliable description of the off - shell behaviour of the effective @xmath2 interaction , the price to be paid however is the non - gauge invariance of the calculated current .
we showed that after repairing gauge invariance in one of the well - known schemes , we find results that are similar to the ones obtained in models that use less sophisticated @xmath2 interactions .
in particular , we verified that the dominant contribution to virtual bremsstrahlung in this energy range comes from the @xmath1-resonance decay . in detail
we found significant differences : the nucleonic contribution is larger than found in models which only include on - shell @xmath2 scattering information .
ceres collaboration , g. agakichiev _
et al . _ ,
report no .
cern ppe/95 - 26 , 1995 ( to be published ) ; j.p .
wurm for the ceres collaboration , in proceedings of quark matter 95 , nucl . phys . *
a590 * , 103c ( 1995 ) . | we present a model for dilepton production of proton - proton collisions using a realist @xmath0-matrix that by incorporating @xmath1 degrees of freedom fits the @xmath2 scattering data up to 2 gev .
the results we find differ in details from earlier work that use less sophisticated interactions but the overall agreement with these calculations is good . |
sensor networks @xcite can be described as a collection of wireless devices with limited computational abilities which are , due to their ad - hoc communication manner , vulnerable to misbehavior and malfunction .
it is therefore necessary to support them with a simple , _ computationally friendly _ protection system . due to the limitations of sensor networks
, there has been an on - going interest in providing them with a protection solution that would fulfill several basic criteria .
the first criterion is the ability of self - learning and self - tuning .
because maintenance of ad hoc networks by a human operator is expected to be sporadic , they have to have a built - in _ autonomous _
mechanism for identifying user behavior that could be potentially damaging to them .
this learning mechanism should itself minimize the need for a human intervention , therefore it should be self - tuning to the maximum extent .
it must also be computationally conservative and meet the usual condition of high detection rate .
the second criterion is the ability to undertake an action against one or several misbehaving users .
this should be understood in a wider context of co - operating wireless devices acting in collusion in order to suppress or minimize the adverse impact of such misbehavior .
such a co - operation should have a low message complexity because both the bandwidth and the battery life are of scarce nature .
the third and last criterion requires that the protection system does not itself introduce new weaknesses to the systems that it should protect .
an emerging solution that could facilitate implementation of the above criteria are artificial immune systems ( ais ) .
ais are based on principles adapted from the human immune system ( his ) @xcite ; the basic ability of his is an efficient detection of potentially harmful foreign agents ( viruses , bacteria , etc . ) .
the goal of ais , in our setting , is the identification of nodes with behavior that could possibly negatively impact the stated mission of the sensor network .
one of the key design challenges of ais is to define a suitable set of efficient genes .
genes form a basis for deciding whether a node misbehaves . they can be characterized as measures that describe a network s performance from a node s viewpoint . given their purpose , they must be easy to compute and robust against deception .
misbehavior in wireless sensor networks can take upon different forms : packet dropping , modification of data structures important for routing , modification of packets , skewing of the network s topology or creating ficticious nodes ( see @xcite for a more complete list ) .
the reason for sensors ( possibly fully controlled by an attacker ) to execute any form of misbehavior can range from the desire to save battery power to making a given wireless sensor network non - functional .
malfunction can also be considered a type of unwanted behavior .
the human immune system is a rather complex mechanism able to protect humans against an amazing set of extraneous attacks .
this system is remarkably efficient , most of the time , in discriminating between _ self _ and _ non - self _ antigens .
a non - self antigen is anything that can initiate an immune response ; examples are a virus , bacteria , or splinter .
the opposite to non - self antigens are self antigens ; self antigens are human organism s own cells .
the process of t - cells maturation in thymus is used as an inspiration for learning in ais .
the maturation of t - cells ( detectors ) in thymus is a result of a pseudo - random process . after a t - cell
is created ( see fig . [
fig : det_gen ] ) , it undergoes a censoring process called _ negative selection_. during negative selection t - cells that bind self are destroyed .
remaining t - cells are introduced into the body .
the recognition of non - self is then done by simply comparing t - cells that survived negative selection with a suspected non - self .
this process is depicted in fig .
[ fig : non_self_det ] .
it is possible that the self set is incomplete , while a t - cell matures ( tolerization period ) in the thymus .
this could lead to producing t - cells that should have been removed from the thymus and can cause an autoimmune reaction , i.e. it leads to _
false positives_. a deficiency of the negative selection process is that alone it is not sufficient for assessing the damage that a non - self antigen could cause .
for example , many bacteria that enter our body are not harmful , therefore an immune reaction is not necessary .
t - cells , actors of the adaptive immune system , require co - stimulation from the innate immune system in order to start acting .
the innate immune system is able to recognize the presence of harmful non - self antigens and tissue damage , and signal this to certain actors of the adaptive immune system .
the random - generate - and - test approach for producing t - cells ( detectors ) described above is analyzed in @xcite .
in general , the number of candidate detectors to the self set size needs to be exponential ( if a matching rule with fixed matching probability is used ) .
another problem is a consistent underfitting of the non - self set ; there exist `` holes '' in the non - self set that are undetectable . in theory , for some matching rules
, the number of holes can be very unfavorable @xcite . in practical terms , the effect of holes depends on the characteristics of the non - self set , representation and matching rule @xcite .
the advantage of this algorithm is its _ simplicity _ and good experimental results in cases when the number of detectors to be produced is fixed and small @xcite .
a review of other approaches to detector computation can be found in @xcite .
a sensor network can be defined in graph theoretic framework as follows : a sensor network is a net @xmath0 where @xmath1 are the set of nodes and edges at time @xmath2 , respectively .
nodes correspond to sensors that wish to communicate with each other .
an edge between two nodes @xmath3 and @xmath4 is said to exist when @xmath3 is within the radio transmission range of @xmath4 and vice versa .
the imposed symmetry of edges is a usual assumption of many mainstream protocols .
the change in the cardinality of sets @xmath1 can be caused by switching on / off one of the sensors , failure , malfunction , removal , signal propagation , link reliability and other factors .
data exchange in a point - to - point ( uni - cast ) scenario usually proceeds as follows : a user initiated data exchange leads to a route query at the network layer of the osi stack .
a routing protocol at that layer attempts to find a route to the data exchange destination .
this request may result in a path of non - unit length .
this means that a data packet in order to reach the destination has to rely on successive forwarding by intermediate nodes on the path .
an example of an on - demand routing protocol often used in sensor networks is dsr @xcite .
route search in this protocol is started only when a route to a destination is needed .
this is done by flooding the network with rreq control packets .
the destination node or an intermediate node that knows a route to the destination will reply with a rrep control packet .
this rrep follows the route back to the source node and updates routing tables at each node that it traverses .
a rerr packet is sent to the connection originator when a node finds out that the next node on the forwarding path is not replaying . at the mac layer of the osi protocol stack , the medium reservation is often contention based . in order to transmit a data packet , the ieee 802.11 mac protocol uses carrier sensing with an rts - cts - data - ack handshake .
should the medium not be available or the handshake fails , an exponential back - off algorithm is used .
this is combined with a mechanism that makes it easier for neighboring nodes to estimate transmission durations .
this is done by exchange of duration values and their subsequent storing in a data structure known as network allocation vector ( nav ) . with the goal to save battery power ,
researchers suggested , a sleep - wake - up schedule for nodes would be appropriate .
this means that nodes do not listen continuously to the medium , but switch themselves off and wake up again after a predetermined period of time .
such a sleep and wake - up schedule is similarly to duration values exchanged among nodes .
an example of a mac protocol , designed specifically for sensor networks , that uses such a schedule is the s - mac @xcite . a sleep and wake - up schedule
can severely limit operation of a node in _ promiscuous mode_. in promiscuous mode , a node listens to the on - going traffic in the neighborhood and collects information from the overheard packets .
this technique is used e.g. in dsr for improved propagation of routing information .
movement of nodes can be modeled by means of a mobility model .
a well - known mobility model is the _ random waypoint model _ @xcite . in this model ,
nodes move from the current position to a new randomly generated position at a predetermined speed . after reaching the new destination
a new random position is computed .
nodes pause at the current position for a time period @xmath2 before moving to the new random position . for more information on sensor networks ,
we refer the reader to @xcite .
motivated by the positive results reported in @xcite we have undertaken a detailed performance study of ais with focus on sensor networks .
the general conclusions that can be drawn from the study presented in this document are : 1 . given the ranges of input parameters that we used and considering the computational capabilities of current sensor devices , we conclude that ais based misbehavior detection offers a decent detection rate . 2 .
one of the main challenges in designing well performing ais for sensor networks is the set of `` genes '' .
this is similar to observations made in @xcite . 3 .
our results suggest that to increase the detection performance , an ais should benefit from information available at all layers of the osi protocol stack ; this includes also detection performance with regards to a simplistic flavor of misbehavior such as packet dropping .
this supports ideas shortly discussed in @xcite where the authors suggest that information available at the application layer deserves more attention .
we observed that somewhat surprisingly a gene based purely on the mac layer significantly contributed to the overall detection performance .
this gene poses less limitations when a mac protocol with a sleep - wake - up schedule such as the s - mac @xcite is used .
it is desirable to use genes that are _ `` complementary '' _ with respect to each other .
we demonstrated that two genes , one that measures correct forwarding of data packets , and the other one that indirectly measures the medium contention , have exactly this property .
we only used a single instance of learning and detection mechanism per node .
this is different from approach used in @xcite , where one instance was used for each of @xmath5 possible neighbors .
our performance results show that the approach in @xcite may not be feasible for sensor networks .
it may allow for an easy sybil attack and , in general , @xmath6 instances might be necessary , where @xmath7 is the total number of sensors in the network . instead , we suggest that flagging a node as misbehaving should , if possible , be based on detection at several nodes .
7 . only less than 5% detectors were used in detecting misbehavior .
this suggests that many of the detectors do not comply with constraints imposed by the communications protocols ; this is an important fact when designing ais for sensor networks because the memory capacity at sensors is expected to be very limited .
the data traffic properties seem not to impact the performance .
this is demonstrated by similar detection performance , when data traffic is modeled as constant bit rate and poisson distributed data packet stream , respectively .
we were unable to distinguish between nodes that misbehave ( e.g. deliberately drop data packets ) and nodes with a behavior resembling a misbehavior ( e.g. drop data packets due to medium contention ) .
this motivates the use of danger signals as described in @xcite .
the approach applied in @xcite does , however , not completely fit sensor networks since these might implement only a simplified version of the transport layer .
in our approach , each node produces and maintains its own set of detectors .
this means that we applied a direct one - to - one mapping between a human body with a thymus and a node .
we represent self , non - self and detector strings as bit - strings .
the matching rule employed is the _ r - contiguous bits matching rule_. two bit - strings of equal length match under the r - contiguous matching rule if there exists a substring of length @xmath8 at position @xmath9 in each of them and these substrings are identical .
detectors are produced by the process shown in fig .
[ fig : det_gen ] , i.e. by means of negative selection when detectors are created randomly and tested against a set of self strings .
each antigen consists of several genes .
_ genes _ are performance measures that a node can acquire locally without the help from another node . in practical terms
this means that an antigen consists of @xmath10 genes ; each of them encodes a performance measure , averaged in our case over a time window .
an antigen is then created by concatenating the @xmath10 genes . when choosing the correct genes , the choice is limited due to the simplified osi protocol stack of sensors . for example
, mica2 sensors @xcite using the tinyos operating system do not guarantee any end - to - end connection reliability ( transport layer ) , leaving only data traffic at the lower layers for consideration .
let us assume that the routing protocol finds for a connection the path @xmath11 from the source node @xmath12 to the destination node @xmath13 , where @xmath14 and @xmath15 .
we have used the following _ genes _ to capture certain aspects of mac and routing layer traffic information ( we averaged over a time period ( window size ) of 500 seconds ) : 1 . *
mac layer : * 2 .
ratio of complete mac layer handshakes between nodes @xmath16 and @xmath17 and rts packets sent by @xmath16 to @xmath17 .
if there is no traffic between two nodes this ratio is set to @xmath18 ( a large number ) .
this ratio is averaged over a time period .
a complete handshake is defined as a completed sequence of rts , cts , data , ack packets between @xmath16 and @xmath17 .
3 . ratio of data packets sent from @xmath16 to @xmath17 and then subsequently forwarded by @xmath17 to @xmath19 . if there is no traffic between two nodes this ratio is set to @xmath18 ( a large number ) .
this ratio is computed by @xmath16 in promiscuous mode and , as in the previous case , averaged over a time period .
this gene was adapted from the watchdog idea in @xcite .
time delay that a data packet spends at @xmath17 before being forwarded to @xmath19 .
the time delay is observed by @xmath16 in promiscuous mode . if there is no traffic between two nodes the time delay
is set to zero .
this measure is averaged over a time period .
this gene is a quantitative extension of the previous gene .
* routing layer : * 6 . the same ratio as in # 2 but computed separately for rerr routing packets
. 7 . the same delay as in # 3 but computed separately for rerr routing packets .
the gene # 1 can be characterized as mac layer quality oriented _ it indirectly measures the medium contention level_. the remaining genes are watchdog oriented .
this means that they more strictly fit a certain kind of misbehavior .
the gene # 2 can help detect whether packets get correctly forwarded ; the gene # 3 can help detect whether forwarding of packets does not get intentionally delayed . as we will show later , in the particular type of misbehavior ( packet dropping ) that we applied ,
the first two genes come out as `` the strongest '' .
the disadvantage of the watchdog based genes is that due to limited battery power , nodes could operate using a sleep - wake - up schedule similar to the one used in the s - mac .
this would mean that the node @xmath16 has to stay awake until the node @xmath17 ( monitored node ) correctly transmits to @xmath19 .
the consequence would be a longer wake - up time and possible restrictions in publishing sleep - wake - up schedules .
in @xcite the authors applied a different a set of genes , based only on the dsr routing protocol .
the observed set of events was the following : a = rreq sent , b = rrep sent , c = rerr sent , d = data sent and ip source address is not of the monitored ( neighboring ) node , e = rreq received , f = rrep received , g = rerr received , h = data received and the ip destination address is not of the monitored node .
the events d and h take into consideration that the source and destination nodes of a connection might appear as misbehaving as they seem to `` deliberately '' create and delete data packets .
then the set of their four genes is as follows :
number of e over a time period .
2 . number of ( e*(a or b ) ) over a time period .
number of h over a time period .
4 . number of ( h*d ) over a time period .
the time period ( window size ) in their case was 10s ; * is the kleene star operator ( zero or more occurrences of any event(s ) are possible ) .
similar to our watchdog genes , these genes impose additional requirements on mac protocols such as the s - mac .
their dependence on the operation in promiscuous mode is , however , more pronounced as a node has to continuously observe packet events at all monitored nodes .
the research in the area of what and to what extent can be or should be locally measured at a node , is independent of the learning mechanism used ( negative selection in both cases ) .
performance of an ais can partly depend on the ordering and the number of used genes .
since longer antigens ( consisting of more genes ) indirectly imply more candidate detectors , the number of genes should be carefully considered .
given @xmath10 genes , it is possible to order them in @xmath20 different ways . in our experience ,
the rules for ordering genes and the number of genes can be summed up as follows : 1 ) keep the number of genes small . in our experiments
, we show that with respect to the learning mechanism used and the expected deployment ( sensor networks ) , 2 - 3 genes are enough for detecting a basic type of misbehavior .
2 ) order genes either randomly or use a predetermined fixed order . defining a utility relation between genes , and ordering genes with respect to it can , in general , lead to problems that are considered intractable .
our results however suggest , it is important to understand relations between different genes , since genes are able to complement each other ; this can lead to their increased mutual strength . on the other hand
, random ordering adds to robustness of the underlying ais . for an attacker ,
it is namely more difficult to deceive , since he does not know how genes are being used .
it is currently an open question , how to impose a balanced solution .
3 ) genes can not be considered in isolation .
our experiments show , when a detector matched an antigen under the @xmath8-contiguous matching rule , usually this match spanned over several genes .
this motivates design of matching rules that would not limit matching to a few neighboring genes , offer more flexibility but still require that a gene remains a partly atomic unit . learning and detection
is done by applying the mechanisms shown in figs .
[ fig : det_gen ] and [ fig : non_self_det ] .
the detection itself is very straightforward . in the learning phase , a misbehavior - free period ( see @xcite on possibilities for circumventing this problem ) is necessary so that nodes get a chance to learn what is the normal behavior . when implementing the learning phase , the designer gets to choose from two possibilities : 1 ) learning and detection at a node get implemented for each neighboring node separately
this means that different antigens have to get computed for each neighboring node , detector computation is different for each neighboring node and , subsequently , detection is different for each neighboring node .
the advantage of this approach is that the node is able to directly determine which neighboring node misbehaves ; the disadvantage is that @xmath5 instances ( @xmath5 is the number of neighbors or node degree ) of the negative selection mechanism have to get executed ; this can be computationally prohibitive for sensor networks as @xmath5 can , in general , be equal to the total number of sensor .
this allows for an easy sybil attack @xcite in which a neighbor would create several identities ; the node would then be unable to recognize that these identities belong to the same neighbor .
this approach was used in @xcite .
2 ) learning and detection at a node get implemented in a single instance for all neighboring nodes .
this means a node is able to recognize anomaly ( misbehavior ) but it may be unable to determine which one from the @xmath5 neighboring nodes misbehaves .
this implies that nodes would have to cooperate when detecting a misbehaving node , exchange anomaly information and be able to draw a conclusion from the obtained information .
an argument for this approach is that in order to detect nodes that misbehave in collusion , it might be necessary to rely to some extent on information exchange among nodes , thus making this a natural solution to the problem .
we have used this approach ; a post - processing phase ( using the list of misbehaving nodes ) was necessary to determine whether a node was correctly flagged as misbehaving or not .
we find the second approach to be more suited for wireless sensor networks .
it is namely less computationally demanding .
we are unable , at this time , to estimate the frequency of a complete detector set computation .
both approaches can be classified within the four - layer architecture ( fig .
[ fig : arch ] ) that we introduced in @xcite .
the lowermost layer , data collection and preprocessing , corresponds to genes computation and antigen construction .
the learning layer corresponds to the negative selection process .
the next layer , local and co - operative detection , suggests , an ais should benefit from both local and cooperative detection .
both our setup and the setup described in @xcite only apply _ local _ detection .
the uppermost layer , local and co - operative response , implies , an ais should also have the capability to undertake an action against one or several misbehaving nodes ; this should be understood in a wider context of co - operating wireless devices acting in collusion in order to suppress or minimize the adverse impact of such misbehavior .
to our best knowledge , there is currently no ais implementation for sensor networks taking advantage of this layer .
_ which @xmath8 is the correct one ? _
an interesting technical problem is to tune the @xmath8 parameter for the @xmath8-contiguous matching rule so that the underlying ais offers good detection and false positives rates .
one possibility is a lengthy simulation study such as this one . through multiparameter simulation
we were to able to show that @xmath21 offers the best performance for our setup . in @xcite
we experimented with the idea of `` growing '' and `` shrinking '' detectors ; this idea was motivated by @xcite .
the initial @xmath22 for a growing detector can be chosen as @xmath23 , where @xmath24 is the detector length .
the goal is to find the smallest @xmath8 such that a candidate detector does not match any self antigen .
this means , initially , a larger ( more specific ) @xmath8 is chosen ; the smallest @xmath8 that fulfills the above condition can be found through binary search . for shrinking detectors
, the approach is reciprocal .
our goal was to show that such growing or shrinking detectors would offer a better detection or false positives rate .
short of proving this in a statistically significant manner , we observed that the growing detectors can be used for self tuning the @xmath8 parameter .
the average @xmath8 value was close to the @xmath8 determined through simulation ( the setup in that case was different from the one described in this document ) .
our experiments show that only a small number of detectors get ever used ( less than 5% ) .
the reason is , they get produced in a random way , not considering structure of the protocols .
for example , a detector that is able to detect whether i ) data packets got correctly transmitted and ii ) 100% of all mac layers handshakes were incomplete is superfluous as this case should never happen . in @xcite ,
the authors conclude : _
`` ... uniform coverage of non - self space is not only unnecessary , it is impractical ; non - self space is too big''_. application driven knowledge can be used to set up a rule based system that would exclude infeasible detectors ; see @xcite for a rule based system aimed at improved coverage of the non - self set . in @xcite , it is suggested that unused detectors should get deleted and the lifetime of useful detectors should be extended . in a companion paper @xcite , we have reviewed different types of misbehavior at the mac , network and transport layers of the osi protocol stack .
we note that solutions to many of these attacks have been already proposed ; these are however specific to a given attack . additionally , due to the limitations of sensor networks , these solutions can not be directly transfered . the appeal of ais based misbehavior detection rests on its simplicity and applicability in an environment that is extremely computationally and bandwidth limited .
misbehavior in sensor networks does not have to be executed by sensors themselves ; one or several computationally more powerful platforms ( laptops ) can be used for the attack . on the other hand , a protection using such more advanced computational platforms is , due to e.g. the need to supply them continuously with electric power , harder to imagine .
it would also create a point of special interest for the possible attackers .
the purpose of our experiments was to show that ais are a viable approach for detecting misbehavior in sensor networks .
furthermore , we wanted to cast light on internal performance of an ais designed to protect sensor networks .
one of our central goals was to provide an in - depth analysis of relative usefulness of genes .
_ definitions of input and output parameters : _ the input parameters for our experiments were : @xmath8 parameter for the @xmath8-contiguous matching rule , the ( desired ) number of detectors and misbehavior level .
misbehavior was modeled as random packet dropping at selected nodes .
+ the performance ( output ) measures were arithmetic averages and 95% confidence intervals @xmath25 of detection rate , number of false positives , real time to compute detectors , data traffic rate at nodes , number of iterations to compute detectors ( number of random tries ) , number of non - valid detectors , number of different ( unique ) antigens in a run or a time window , and number of matches for each gene .
the detection rate @xmath26 is defined as @xmath27 , where @xmath28 is the number of detected non - self strings and @xmath29 is the total number of non - self strings . a false positive in our definition
is a string that is not self but can still be a result of anomaly that is identical with the effects of a misbehavior .
a non - valid detector is a candidate detector that matches a self string and must therefore be removed .
the number of matches for each gene was evaluated using the @xmath8-contiguous matching rule ; we considered two cases : i ) two bit - strings get matched from the left to the right and the first such a match will get reported ( matching gets interrupted ) , ii ) two bit - strings get matched from the left to the right and all possible matches will get reported .
the time complexity of these two approaches is @xmath30 and @xmath31 , respectivelly ; @xmath32 , where @xmath24 is the bitstring length .
the first approach is exactly what we used when computing the real time necessary for negative selection , the second approach was used when our goal was to evaluate relative usefulness of each gene .
* scenario description : * we wanted to capture `` self '' and `` non - self '' packet traffic in a large enough synthetic static sensor network and test whether using an ais we are able to recognize non - self , i.e. misbehavior .
the topology of this network was determined by making a _ snapshot _ of 1,718 mobile nodes ( each with 100 m radio radius ) moving in a square area of 2,900m@xmath332,950 m as prescribed by the random waypoint mobility model ; see figure [ fig : topo](a ) .
the motivation in using this movement model and then creating a snapshot are the results in our previous paper @xcite that deals with structural robustness of sensor network .
our preference was to use a slightly bigger network than it might be necessary , rather than using a network with unknown properties .
the computational overhead is negligible ; simulation real time mainly depends on the number of events that require processing .
idle nodes increase memory requirements , but memory availability at computers was in our case not a bottleneck .
we chose source and destination pairs for each connection so that several alternative independent routes exist ; the idea was to benefit from route repair and route acquisition mechanisms of the dsr routing protocol , so that the added value of ais based misbehavior detection is obvious .
we used 10 cbr ( constant bit rate ) connections .
the connections were chosen so that their length is @xmath347 hops and so that these connections share some common intermediate nodes ; see figure [ fig : topo](b ) . for each packet received or sent by a node
we have captured the following information : ip header type ( udp , 802.11 or dsr in this case ) , mac frame type ( rts , cts , data , ack in the case of 802.11 ) , current simulation clock , node address , next hop destination address , data packet source and destination address and packet size . _
encoding of self and non - self antigens : _ each of the five genes was transformed in a 10-bit signature where each bit defines an interval of a gene specific value range .
we created self and non - self antigen strings by concatenation of the defined genes .
each self and non - self antigen has therefore a size of 50 bits .
the interval representation was chosen in order to avoid carry - bits ( the gray coding is an alternative solution ) .
_ constructing the self and non - self sets : _ we have randomly chosen 28 non - overlapping 500-second windows in our 4-hour simulation . in each 500-second window self and non - self antigens
are computed for each node .
this was repeated 20 times for independent glomosim runs . _
misbehavior modeling : _ misbehavior is modeled as random data packet dropping ( implemented at the network layer ) ; data packets include both data packets from the transport layer as well as routing protocol packets . that should get dropped will simply not be inserted into the ip queue ) ; we have randomly chosen 236 nodes and these were forced to drop @xmath35 of data packets .
however , there were only 3 - 10 nodes with misbehavior and with a _ statistically significant _ number of packets for forwarding in each simulation run ; see constraint c2 in section [ sec : eval ] .
_ detection : _ a neighboring node gets flagged as misbehaving , if a detector from the detector set matches an antigen .
since we used a single learning phase , we had to complement this process with some routing information analysis .
this allowed us to determine , which one from the neighboring nodes is actually the misbehaving one . in the future
, we plan to rely on co - operative detection in order to replace such a post - analysis .
_ simulation phases : _ the experiment was done in four phases . 1 .
20 independent glomosim runs were done for one of @xmath35 misbehavior levels and `` normal '' traffic .
normal means that no misbehavior took place .
2 . self and non - self antigen computation ( encoding ) .
the 20 `` normal '' traffic runs were used to compute detectors . given the 28 windows and 20 runs , the sample size was 20@xmath3328 = 560 , i.e. detectors at each node were discriminated against 560 self antigens .
4 . using the runs with @xmath35 misbehavior levels ,
the process shown in fig .
[ fig : non_self_det ] was used for detection ; we restricted ourselves to nodes that had in both the normal and misbehavior traffic at least a certain number of data packets to forward ( packet threshold ) .
the experiment was then repeated with different @xmath8 , desired number of detectors and misbehavior level .
the parameters for this experiment are summarized in fig .
[ fig : exp - parameters2 ] .
the injection rate and packet sizes were chosen in order to comply with usual data rates of sensors ( e.g. 38.4kbps for mica2 ; see @xcite ) .
we chose the glomosim simulator @xcite over other options ( most notably ns2 ) because of its better scaling characteristics @xcite and our familiarity with the tool .
when evaluating our results we define two additional constraints : 1 .
we define a node to be detected as misbehaving if it gets flagged in at least 14 out of the 28 possible windows .
this notion indirectly defines the time until a node is pronounced to be misbehaving .
we call this a _ window threshold_. 2
. a node @xmath16 has to forward in average at least @xmath5 packets over the 20 runs in both the `` normal '' and misbehavior cases in order to be included into our statistics .
this constraint was set in order to make the detection process more reliable .
it is dubious to flag a neighboring node of @xmath16 as misbehaving , if it is based on `` normal '' runs or runs with misbehavior , in which node @xmath16 had no data packets to forward ( he was not on a routing path ) .
we call this a _ packet threshold _ ; @xmath5 was in our simulations chosen from @xmath36 .
_ example : _ for a fixed set of input parameters , a node forwarded in the `` normal '' runs in average 1,250 packets and in the misbehavior runs ( with e.g. level 30% ) 750 packets .
the node @xmath16 would be considered for misbehavior detection if @xmath37 , but not if @xmath38 . in other words ,
a node has to get a chance to learn what is `` normal '' and then to use this knowledge on a non - empty packet stream .
the results related to computation of detectors are shown in figure [ fig:300 m ] . in our experiments
we have considered the desired number of detectors to be max .
4,000 ; over this threshold the computational requirements might be too high for current sensor devices .
we remind the reader , each time the @xmath8 parameter is incremented by @xmath39 , the number of detectors should double in order to make these two cases comparable .
figure [ fig:300m](a ) shows the real time needed to compute the desired set of detectors .
we can see the real time necessary increases proportionally with the desired number of detectors ; this complies with the theoretical results presented in @xcite .
figure [ fig:300m](b ) shows the percentage of non - valid detectors , i.e. candidate detectors that were found to match a self string ( see figure [ fig : det_gen ] ) .
this result points to where the optimal operation point of an ais might lie with respect to the choice of @xmath8 parameter and the choice of a fixed number of detectors to compute .
we remind the reader , the larger is the @xmath8 parameter the smaller is the probability that a detector will match a self string .
therefore overhead connected to choosing the @xmath8 parameter prohibitively small should be considered when designing an ais .
figure [ fig:300m](c ) shows the total number of generate - and - test tries needed for computation of detector set of a fixed size ; the 95% confidence interval is less than @xmath40 . in figure
[ fig : perform](a ) we show the dependence of detection ratio on the packet threshold .
we conclude that except for some extremely low threshold values ( not shown ) the detection rate stays constant .
this figure also shows that when misbehavior level was set very low , i.e. 10% , the ais struggled to detect misbehaving nodes .
this is partly a result of our coarse encoding with only 10 different levels . at the 30 and 50% misbehaving levels the detection rate stays solid at about 70 - 85% .
the range of the 95% confidence interval of detection rate is 3.8 - 19.8% .
the fact that the detection rate did not get closer to 100% suggests , either the implemented genes are not sufficient , detection should be extended to protocols at other layers of the osi protocol stack , a different ordering of genes should have been applied or our ten level encoding was too coarse .
it also implicates that watchdog based genes ( though they perfectly fit the implemented misbehavior ) should not be used in isolation , and in general , that the choice of genes has to be very careful
. figure [ fig : perform](b ) shows the impact of @xmath8 on detection rate . when @xmath41 the ais performs well , for @xmath42 the detection rate decreases .
this is caused by the inadequate numbers of detectors used at higher levels of @xmath8 ( we limited ourselves to max .
4,000 detectors ) . figure [ fig : perform](c ) shows the number of false positives .
we remind that in our definition false positives are both nodes that do not drop any packets and nodes that drop packets due to other reasons than misbehavior . in a separate experiment we studied whether the 4-hour ( 560 samples ) simulation time was enough to capture the diversity of the self behavior .
this was done by trying to detect misbehavior in 20 independent misbehavior - free glomosim runs ( different from those used to compute detectors ) .
we report that we did not observe a single case of an autoimmune reaction . in fig .
[ fig : det_used1](a ) we show the total number of runs in which a node was identified as misbehaving . the steep decline for values @xmath43 ( in this and other figures ) documents that in these cases it was necessary to produce a higher number of detectors in order to cover the non - self antigen space .
the higher the @xmath8 , the higher is the specificity of a detector , this means that it is able to match a smaller set of non - self antigens . in fig .
[ fig : det_used1](b ) and ( c ) we show the number of detectors that got matched during the detection phase ( see fig .
[ fig : non_self_det ] ) . fig .
( b ) shows the number of detectors matched per run , fig .
( c ) shows the number of detectors matched per window .
( b ) is an upper estimate on the number of unique detectors needed in a single run .
given that the total number of detectors was 2,000 , there were less than 5% detectors that would get used in the detection phase .
the tight confidence intervals only for @xmath44 . ] for the number of unique detectors matched per window ( see fig .
( c ) ) is a direct consequence of the small variability of antigens as shown in fig .
[ fig : det_used2](a ) . fig .
[ fig : det_used2](a ) shows the number of unique antigens that were subject to classification into self or non - self .
the average for @xmath45 is about 1.5 .
this fact does not directly imply that the variability of the data traffic would be inadequate .
it is rather a direct consequence of our choice of genes and their encoding ( we only used 10 value levels for encoding ) .
[ fig : det_used2](b ) shows the number of matches between a detector and an antigen in the following way .
when a detector under the @xmath8-contiguous matching rule matches only a single gene within an antigen , we would increment the `` single '' counter .
otherwise , we would increment the `` multiple '' counter .
it is obvious that with increasing @xmath8 , it gets more and more probable that a detector would match more than a single gene .
the interesting fact is that the detection rate for both @xmath46 and @xmath47 is about 80% ( see fig .
[ fig : perform](a ) ) and that the rate of non - valid detectors is very different ( see fig .
[ fig:300m](b ) ) .
this means that an interaction between genes has positively affected the later performance measure , without sacrificing on the former one .
this leads to a conlusion that genes should not be considered in isolation . fig .
[ fig : det_used2](c ) shows the performance of gene # 1 .
the number of matches shows that this gene contributed to the overall detection performance of our ais .
[ fig : det_used3](a - c ) sum up performance of the five genes for different values of @xmath8 .
again , an interesting fact is the contribution of gene # 1 to the overall detection performance .
the usefulness of gene # 2 was largely expected as this gene was tailored for the kind of misbehavior that we implemented .
the other three genes came out as marginally useful .
the importance of the somewhat surprising performance of gene # 1 is that it can be computed in a simplistic way and does not require continuous operation of a node . in an additional experiment
, we examined the impact of data traffic pattern on the performance .
we used two different data traffic models : the constant bit rate ( cbr ) and a poisson distributed data traffic . in many scenarios ,
sensors are expected to take measurements in constant intervals and , subsequently , send them out for processing .
this would create a constant bit rate traffic .
poisson distributed traffic could be a result of sensors taking measurements in an event - driven fashion .
for example , a sensor would take a measurement only when a target object ( e.g. a person ) happens to be in its vicinity .
the setup for this experiment was similar to that presented in fig .
[ fig : exp - parameters2 ] with the additional fact that the data traffic model would now become an input parameter . with the goal to reduce complexity of the experimental setup , we fixed @xmath47 and we only considered cases with @xmath48 and @xmath49 detectors . in order to match the cbr traffic rate , the poisson distributed data traffic model had a mean arrival expectation of @xmath39 packet per second ( @xmath50 ) .
as in the case with cbr , we computed the detection rate and the rate of false positives with the associated arithmetic averages and @xmath51 confidence intervals . the results based on these two traffic models were similar , actually , we could not find the difference between them to be statistically significant .
this points out that the detection process is robust against some variation in data traffic .
this conclusion also reflects positively on the usefulness of the used genes .
more importantly , it helped disperse our worries that the results presented in this experimental study could be unacceptably data traffic dependent .
in @xcite the authors introduced an ais based misbehavior detection system for ad hoc wireless networks .
they used glomosim for simulating data traffic , their setup was an area of 800@xmath33600 m with 40 mobile nodes ( speed 1 m / s ) of which 5 - 20 are misbehaving ; the routing protocol was dsr .
four genes were used to capture local behavior at the network layer .
the misbehavior implemented is a subset of misbehavior introduced in this paper ; their observed detection rate is about 55% .
additionally , a co - stimulation in the form of a danger signal was used in order to inform nodes on a forwarding path about misbehavior , thus propagating information about misbehaving nodes around the network . in @xcite
the authors describe an ais able to detect anomalies at the transport layer of the osi protocol stack ; only a wired tcp / ip network is considered .
self is defined as normal pairwise connections .
each detector is represented as a 49-bit string .
the pattern matching is based on r - contiguous bits with a fixed @xmath52 .
@xcite discusses a network intrusion system that aims at detecting misbehavior by capturing tcp packet headers .
they report that their ais is unsuitable for detecting anomalies in communication networks .
this result is questioned in @xcite where it is stated that this is due to the choice of problem representation and due to the choice of matching threshold @xmath8 for @xmath8-contiguous bits matching . to overcome the deficiencies of the generate - and - test approach a different approach
is outlined in @xcite .
several signals each having a different function are employed in order to detect a specific misbehavior in sensor wireless networks .
unfortunately , no performance analysis was presented and the properties of these signals were not evaluated with respect to their misuse .
the main discerning factor between our work and works shortly discussed above is that we carefully considered hardware parameters of current sensor devices , the set of input parameters was designed in order to target specifically sensor networks and our simulation setup reflects structural qualities of such networks with regards to existence of multiple independent routing paths . in comparison to @xcite we showed that in case of static sensor networks it is reasonable to expect the detection rate to be above 80% .
although we answered some basic question on the suitability and feasibility of ais for detecting misbehavior in sensor networks a few questions remain open .
the key question in the design of ais is the quantity , quality and ordering of genes that are used for measuring behavior at nodes . to answer this question a detailed formal analysis of communications protocols
will be needed .
the set of genes should be as `` complete '' as possible with respect to any possible misbehavior .
the choice of genes should impose a high degree of sensor network s survivability defined as _the capability of a system to fulfill its mission in a timely manner , even in the presence of attacks , failures or accidents _
it is therefore of paramount importance that the sensor network s mission is clearly defined and achievable under normal operating conditions .
we showed the influence and usefulness of certain genes in order to detect misbehavior and the impact of the @xmath8 parameter on the detection process .
in general , the results in fig . [
fig : det_used3 ] show that gene # 1 and # 2 obtained of all genes the best results , with gene # 2 showing always the best results .
the contribution of gene # 1 suggests that observing the mac layer and the ratio of complete handshakes to the number of rts packets sent is useful for the implemented misbehaviour .
gene # 2 fits perfectly for the implemented misbehavior .
it therefore comes as no surprise that this gene showed the best results in the detection process .
the question which remains open is whether the two genes are still as useful when exposed to different attack patterns .
it is currently unclear whether genes that performed well with negative selection , will also be appropriate for generating different flavors of signals as suggested within the _ danger theory _
it is our opinion that any set of genes , whether used with negative selection or for generating any such a signal , should aim at capturing intrinsic properties of the interaction among different components of a given sensor network .
this contradicts approaches applied in @xcite where the genes are closely coupled with a given protocol .
the reason for this statement is the _ combined performance _ of gene # 1 and # 2 .
their interaction can be understood as follows : data packet dropping implies less medium contention since there are less data packets to get forwarded .
less data packets to forward on the other hand implies easier access to the medium , i.e. the number of complete mac handshakes should increase .
this is an interesting _ complementary _ relationship since in order to deceive these two genes , a misbehaving node has to appear to be correctly forwarding data packets and , at the same time , he should not significantly modify the `` game '' of medium access .
it is improbable that the misbehaving node _ alone _ would be able to estimate the impact of dropped packets on the contention level .
therefore , he lacks an important feedback mechanism that would allow him to keep the contention level unchanged . for that
, he would need to act in collusion with other nodes .
the property of complementarity moves the burden of excessive communication from normally behaving nodes to misbehaving nodes , thus , exploiting the ad hoc ( local ) nature of sensor networks .
our results thus imply , _ a `` good '' mixture of genes should be able to capture interactions that a node is unable to influence when acting alone . _
it is an open question whether there exist other useful properties of genes , other than complementarity .
we conclude that the random - generate - and - test process , with no knowledge of the used protocols and their behavior , creates many detectors which might show to be superfluous in detecting misbehavior .
a process with some basic knowledge of protocol limitations might lead to improved quality of detectors . in
@xcite the authors stated that the random - generate - and - test process _
`` is innefficient , since a vast number of randomly generated detectors need to be discarded , before the required number of the suitable ones are obtained '' . _
our results show that at @xmath21 , the rate of discarded detectors is less than @xmath53 .
hence , at least in our setting we could not confirm the above statement .
a disturbing fact is , however , that the size of the self set in our setting was probably too small in order to justify the use of negative selection .
a counter - balancing argument is here the realistic setup of our simulations and a decent detection rate .
we would like to point out that the fisher iris and biomedical data sets , used in @xcite to argue about the apropriateness of negative selection for anomaly detection , could be very different from data sets generated by our simulations .
our experiments show that anomaly ( misbehavior ) data sets based on sensor networks could be in general very sparse .
this effect can be due to the limiting nature of communications protocols .
since the fisher iris and biomedical data sets were in @xcite not evaluated with respect to some basic properties e.g. degree of clustering , it is hard to compare our results with the results presented therein . in order to understand the effects of misbehavior better ( e.g. the propagation of certain adverse effects )
, we currently develop a general framework for ais to be used within the jist / swans network simulator @xcite .
this work was supported by the german research foundation ( dfg ) under the grant no .
sz 51/24 - 2 ( survivable ad hoc networks sane ) . | a sensor network is a collection of wireless devices that are able to monitor physical or environmental conditions . these devices ( nodes )
are expected to operate autonomously , be battery powered and have very limited computational capabilities .
this makes the task of protecting a sensor network against misbehavior or possible malfunction a challenging problem . in this document
we discuss performance of artificial immune systems ( ais ) when used as the mechanism for detecting misbehavior .
we show that ( i ) mechanism of the ais have to be carefully applied in order to avoid security weaknesses , ( ii ) the choice of genes and their interaction have a profound influence on the performance of the ais , ( iii ) randomly created detectors do not comply with limitations imposed by communications protocols and ( iv ) the data traffic pattern seems not to impact significantly the overall performance .
we identified a specific mac layer based gene that showed to be especially useful for detection ; genes measure a network s performance from a node s viewpoint .
furthermore , we identified an interesting complementarity property of genes ; this property exploits the local nature of sensor networks and moves the burden of excessive communication from normally behaving nodes to misbehaving nodes .
these results have a direct impact on the design of ais for sensor networks and on engineering of sensor networks .
= 1.0 sensor networks , ad hoc wireless networks , artificial immune systems , misbehavior . |
the axion@xcite is the nambu - goldstone boson associated with the peccei - quinn symmetry breaking which was invented as the most natural solution to the strong cp problem@xcite of qcd .
the peccei - quinn symmetry breaking scale @xmath1 is stringently constrained by the consideration of accelerator experiments , stellar cooling and cosmology .
the allowed range of @xmath2 ( axion window ) is between @xmath3 and @xmath4 , where the integer @xmath5 is the color anomaly of peccei - quinn symmetry . since the peccei - quinn symmetry is a global u(1 ) symmetry , global strings ( axionic strings )
are produced during spontaneous symmetry breaking . at this stage
the potential for the axion field is flat , i.e. , the axion is massless .
however the axion has a mass at qcd scale through the instanton effect .
the potential is written by @xmath6 where @xmath7 is the pion decay constant and @xmath8 is its mass .
then the mass of the axion is given by @xmath9 .
this potential([potential ] ) has @xmath10 symmetry and takes its minimum at @xmath11 .
@xmath10 discrete symmetry is spontaneously broken to produce two dimensional topological defects , i.e. , axionic domain walls .
the property of the axionic domain wall is characterized by @xmath5 .
since @xmath5 domain walls stretch out from each axionic string , the network of the string - wall systems is very complicated and survives long enough to dominate the universe for @xmath12 axionic domain wall @xcite .
therefore the @xmath13 domain wall is not accepted cosmologically .
one might expect that the density of the domain wall can be diluted in the inflationary universe . however , for the dilution mechanism to work , both the reheating temperature and the expansion rate during inflation should be smaller than @xmath1 @xcite , which is rather unnatural requirement for many inflation models .
therefore the domain wall problem in the axion model is serious one . in the case of @xmath14 ,
the domain wall is a disk bounded by the axionic string .
the wall with the string boundary is no longer stable and it might collapse by the surface tension .
in fact , since , as seen later , the surface tension is stronger than the tension due to the string for the domain wall with size much greater than the width of the wall , the dynamics of the wall is controlled by the wall tension .
therefore the string that bounds the wall can not prevent the wall from collapsing . typically the wall has size of horizon length ( @xmath15 ) when it is formed .
therefore the time scale for collapse is about @xmath16 sec and the walls disappear quickly without overclosing the universe and hence the domain wall seems harmless in the case of @xmath14 axion model .
however it is expected that a number of axions are produced when the axionic walls collapse .
if produced axions are non - relativistic or cold , their energy density decreases slowly ( @xmath17 @xmath18 : scale factor ) compared with the radiation density ( @xmath19 ) .
hence the relative contribution of axions to the total density of the universe increases with @xmath18 until the universe becomes matter dominated .
in this paper we study the collapse of the axionic domain wall ( @xmath14 ) by the numerical integration of equations of motion for the axion field and estimate the energy density of axion field after the collapse .
it is found that the energy carried by the walls is converted to the axions , which gives significant contribution to the mass density of the present universe and might account for the dark matter of the universe .
the axion emission is also expected from axionic string before @xmath20@xcite or annihilation of axionic domain walls@xcite . however , since it is shown later that parallel two domain walls go through each other without annihilation , it is unlikely that a large number of axions are produced by interaction of two domain walls .
the massless axion can be produced by axionic strings and they acquire mass after the qcd phase transition , which might give significant contribution to the total density of the present universe .
however there are two independent quantitative estimations by davis@xcite and by harari and sikivie@xcite and , unfortunately , they are quite different .
the contribution from axionic domain walls is expected to be at least comparable to that from axionic strings if the estimation by harari and sikivie is correct .
the dynamics of the peccei - quinn scalar field @xmath21 is described by the lagrangian : @xmath22 ^ 2 \nonumber\\ & & = \partial^{\mu } \phi^ { * } \partial_{\mu } \phi + v_s(|\phi| ) \label{eq : lagra}\end{aligned}\ ] ] where @xmath23 is the coupling constant .
as the universe cools down , peccei - quinn u(1 ) symmetry ( @xmath24 ) is spontaneously broken and the scalar field has vacuum expectation value @xmath25 .
after global u(1 ) symmetry is broken , axionic strings are formed . from causality argument ,
about one axionic string is produced within the horizon at @xmath26 .
since the line energy density of the string is @xmath27 , the string density @xmath28 is about @xmath29 which is only @xmath30 of total density of the universe at the formation epoch . defining @xmath31 ,
the lagrangian for axion field @xmath32 is derived from eq.([eq : lagra ] ) as @xmath33 where the second term comes from qcd instanton effect which gives axion mass @xmath34 .
the potential @xmath35 has a minimum at @xmath36 and domain walls are produced between @xmath37 phase and @xmath38 phase .
more precisely , the axion mass has temperature dependence and increase as @xmath39 where @xmath40 is the qcd scale @xmath41 .
the domain walls are produced when the axion mass becomes greater than the expansion rate of the universe , i.e. , @xmath42 .
the axionic domain wall has size of about horizon length at the formation time @xmath43 and has the axionic string on its boundary .
the surface tension of the axionic domain wall is @xmath44@xcite . for wall with size @xmath45
, the ratio of surface energy to string energy is given by @xmath46 therefore the dynamics of the axionic wall with size greater than @xmath47 is determined by the wall tension and the string tension can be neglected .
the cosmological evolution of the wall is determined by the surface tension and interaction of walls ; the former makes the wall shrink and the latter cuts the wall into small pieces . in both cases the wall finally collapses by the surface tension .
after collapse the domain walls disappear and the energy that the wall had is converted into kinetic energy of axion fields . if the energy of axion fields changes like non - relativistic particles as the universe expands , the axion energy density is @xmath48 where @xmath49 and @xmath50 are the energy density of the axion field and the axionic wall , respectively . for relativistic axion ,
the axion density decreases as @xmath51 until axion becomes non - relativistic
. then eq.([axiondensity ] ) is changed to @xmath52 where @xmath53 is the average energy of emitted axion at @xmath54 . assuming the mean distance and the mean radius of walls are @xmath55 and @xmath56 , @xmath50 is given by @xmath57 where @xmath58 is the relativistic degrees of freedom at @xmath59 .
we expect that the numerical parameters @xmath60 and @xmath61 are o(1 ) from causality , although the precise values should be determined by the realistic simulation of cosmological formation of the wall - string system , which is beyond the scope of the present paper . using the entropy density of the universe @xmath62 and its conservation ( @xmath63 const . )
, the number density of the axion can be written as @xmath64 then the present density of the axion is @xmath65 since @xmath66 , @xmath67 and @xmath68 , @xmath69 the contribution of the axion to the density parameter @xmath70 is given by @xmath71 where @xmath72 is the hubble constant in units of @xmath73 . therefore the axion density is large enough to account for the dark matter in the universe unless the axion is ultra - relativistic when it is emitted .
we estimate the @xmath74 by numerical simulation in the next section .
in order to follow the motion of domain walls , we have solved the evolution equation of the axion field numerically . when a wall piece much smaller than the cosmological horizon is considered , the cosmic expansion can be ignored .
then the field equation under the minkowski background is written as @xmath75 using the potentials in the equations ( [ eq : lagra ] ) and ( [ eq : alagra ] ) .
we have employed the staggered leapfrog method to solve the differential equation .
the model parameters are chosen such that the width of the wall is equal to ten simulation meshes and the vacuum energy of the string is a hundred times larger than that of the wall .
the variation of the numerical value in the latter condition does not alter our conclusion since only the wall tension governs the motion of walls as we have mentioned above .
the boundary condition is periodic , which is useful to check the accuracy of our calculation since the total energy in the simulation box is conserved .
the solution of a static infinite planar wall under @xmath76 is used in the initial configuration .
when the wall lies in @xmath77 plane , it is expressed as @xmath78 thus the scale of the inverse axion mass represents the characteristic thickness of the axionic wall .
the basic numerical technique is the same one in the previous paper .
see the reference @xcite for more details .
first we have confirmed that approaching walls that face in parallel each other pass through one another .
widrow showed this is true in the case of a toy sine - gordon potential@xcite .
we have reproduced the passing phenomenon by numerical simulations using the potential ( [ eq : eveq ] ) .
fig.[fig : pass ] shows the result .
the initial condition is set so that the relative velocity is @xmath79 , where @xmath80 is light velocity and the separation between walls is 200 meshes . notice that the size of one mesh is equal to @xmath81 in our simulation and we take @xmath82 . as the time evolves ,
two walls become close and go away without any crush . when the initial relative velocity is much larger , for example , equal to @xmath83 ,
the pair annihilation of walls occurs neither .
therefore it is unlikely that annihilation of domain walls occurs and produces axions during the collision process .
the possibility which is described in the reference @xcite should be unfavorable .
on the other hand , in the case of the encounter of a wall with a wall edge , i.e. , a string cuts the wall and the process of disintegration advances@xcite .
hence by repeated intercommutations , large walls are broken to small pieces .
when their size becomes comparable with the thickness scale of the wall , they collapse and radiate energy as axions . in order to see how fast and effective the wall collapse is
, we have performed the simulation of the evolution of a small wall piece . as a simple example we followed the time evolution of a disk wall surrounded by a circular string in the previous work @xcite .
the result shows that the disk wall shrinks at the velocity of light and the energy of the wall is converted to axions .
we can say that the wall collapse process is so rapid that the radiation of gravitational waves is hardly expected . in order to see the wall collapse with higher accuracy
, we have performed the simulations of a strip wall .
figs.[fig : obi](a)(b)(c ) demonstrate the two - dimensional distribution of @xmath84 .
the first one shows the initial configuration in which there is one strip wall of infinite height and 20 meshes length .
as the time evolves , the edges of the strip approach each other , that is , the wall size decreases . thus the wall collapse proceeds also at about the light velocity and the result of the three - dimensional simulation is confirmed . in the case of fig.[fig : obi ] , the initial @xmath85 is zero everywhere in the box .
however , the initial motion may make the wall move periodically so that the death of wall might be avoided .
we have found that this is not the case by performing the simulations with various initial motions . even in the most extreme case where the strings attached to the wall edges go away each other with the light velocity ,
such a motion could do nothing more than the slight extension of the wall life time .
finally we have estimated the @xmath74 in the simulation .
the time evolution of the potential energy @xmath84 and kinetic energy of the axion field @xmath32 is shown in fig.[fig : pot](a ) . in these simulations ,
the wall edges are smoothly connected to the true vacuum region which is different from the disk wall case . since the overestimation of the field gradient is removed by this smoothing
, the quantitative analysis of energy distribution is enabled .
after the wall collapse there remains only axionic waves which we identify as axions and the values of the potential and kinetic energy of the axion become constant .
since the axion waves oscillate like @xmath86 , the ratio of the kinetic energy to potential energy is equal to @xmath87 .
the simulation shows @xmath88 , which means that the emitted axion is mildly relativistic and becomes non - relativistic soon by cooling due to the cosmic expansion .
as mentioned before we check the accuracy of our simulation by the conservation of the total energy in the simulation box . in the simulation above the total energy
is conserved within accuracy of 20% . in order to confirm the estimated value of @xmath89 we have performed the simulation of higher resolution ( string width @xmath90 meshes , wall width @xmath91 meshes , wall length @xmath92 meshes ) . the result ( fig.[fig : pot](b ) )
shows @xmath93 which is quite consistent with the result of the lower resolution run .
our model parameters in the simulations above correspond to @xmath94 which is much larger than the actual axion model ( @xmath95 ) . to check
if the estimation for @xmath74 depends on @xmath96 , we have run the simulation of thicker wall ( string width @xmath97 meshes wall width @xmath98 meshes , wall length @xmath99 meshes ) in which @xmath100 .
although the conservation of the total energy becomes as bad as 50% in this case , the result indicates that the value of @xmath74 is affected only slightly , which means the generality of our estimation concerning the relativisity of emitted axions .
in the previous section , it is found that the axionic wall collapses with time scale @xmath101 and the mildly - relativistic axions remain after the collapse .
the estimated @xmath74 is about 3 , which leads to the relic density of the axion given by @xmath102 since @xmath60 and @xmath61 are expected to be @xmath103 , the contribution of the axion to the present universe is large for @xmath104 ( compared with the baryon density @xmath105@xcite ) and might account for the dark matter of the universe
. however eq.([omega - axion2 ] ) should be compared with the density of the coherent oscillation of axion field ( cold axion)@xcite given by @xmath106 therefore the axion from the wall has comparable to or higher density than the cold axion unless @xmath107 is much less than 1 . in deriving ( [ omega - axion2 ] ) , we assume that the domain wall collapses rapidly after its formation . as shown in the simulations this assumption is quite reasonable for walls whose size is much smaller than the horizon . however
, a large wall does not shrink until its size becomes smaller than the horizon , which leads to the increase of the density of axionic domain wall and as a result the density of the axion produced by the collapse of the wall
. therefore eq.([omega - axion2 ] ) may underestimate the actual axion density and the axion from axionic walls may be more important than the cold axion .
another important source of axions is an oscillating axionic string .
let us compare our result with the axion emission from the axionic strings .
the density of the axion from the strings is given by @xmath108 where uncertainty of a factor of 100 is due to two different estimations by davis@xcite and harari and sikivie@xcite .
thus the importance of the axion from the wall depends on which estimation is correct .
the difference comes from the different assumption for the energy spectrum of the emitted axion and it is hard to judge which assumption is better . in the case of the axionic domain wall
the spectrum or average energy of emitted axions can be estimated by numerical simulations more easily since axions are mostly produced at the final stage of the collapsing wall whose size is much smaller than the cosmic scale . in conclusion , the relic density of the axion produced by the collapse of axionic walls is , at least , larger than the baryon density for @xmath109 gev and accounts for a part or all of the dark matter in the universe .
the axion from the axionic domain wall is more important than the cold axion and comparable to that from strings if harari and sikivie s estimation is correct . to make a more precise prediction of the relic density we have to know the numeric parameter @xmath60 and @xmath61 , which is beyond the scope of the present work and
will be studied in future work .
0.5 cm mn is grateful to professor k. sato for his continuous encouragement .
this work was in part supported by the japanese grant in aid for science research fund of the ministry of education , science and culture ( no .
numerical computations were partially performed by sun sparc stations at uji research center , yukawa institute for theoretical physics , kyoto university .
peccei and h.r .
quinn , phys .
38 * , 1440 ( 1977 ) .
f. wilczeck , phys .
* 40 * , 279 ( 1978 ) .
kim , phys .
* 43 * , 103 ( 1979 ) ; m. shifman , a. vainshtein and v. zakharov , nucl . phys .
* b166 * , 493 ( 1980 ) .
m. dine , w. fischler and m. srednicki , phys .
* b104 * , 199 ( 1981 ) . g. t hooft , phys . rev .
* 37 * , 8 ( 1976 ) .
b. s. ryden , w. h. press and d. n. spergel , ap . j. * 357 * , 293 ( 1990 ) .
a. linde , phys . lett . * b259 * , 38 ( 1991 ) .
r. davis , phys .
b180 * , 225 ( 1986 ) .
d. harari and p. sikivie , phys . lett . *
b195 * , 361 ( 1987 ) ; + c. hagmann and p. sikivie , nucl . phys . *
b363 * , 247 ( 1991 ) d.h .
lyth , phys . lett . *
b275 * , 279 ( 1992 ) .
a. virenkin , phys . rep .
* 121 * , 263 ( 1985 ) .
l. m. widrow , phys . rev . *
d40 * , 1002 ( 1989 ) .
m. nagasawa , phys .
b318 * , 53 ( 1993 ) .
t.p . walker , g. steigman , d.n .
schramm , k.a .
olive and h - s .
kang , astrophys .
j. , * 376 * , 51 ( 1991 ) j. preskill , m. wise and f. wilczek , phys . lett . *
b180 * , 225 ( 1983 ) ; + l. abbott and p. sikivie , phys . lett.*b180
* , 133 ( 1983 ) ; + m. dine and w. fischer , phys .
lett.*b180 * , 137 ( 1983 ) .
turner , phys . rev . *
d33 * , 889 ( 1986 ) . | we examine the collapse of an axion domain wall bounded by an axionic string . it is found that the collapse proceeds quickly and axion
domain walls disappear
. however axions are emitted in the collapse and its energy density increases during radiation dominated era and contributes significantly to the present mass density of the universe . in particular
the axion emitted from the wall can account for the dark matter in the universe for @xmath0 . |
figure [ figs1 ] shows representative histograms of the height values on the sio@xmath0 substrate and the graphene sample as obtained by atomic force microscopy ( afm ) and scanning tunneling microscopy ( stm ) , respectively .
the histograms are based on images of 200 @xmath24 200 nm@xmath7 including 512 @xmath24 512 measurement points each .
these images are shown in fig .
3 of the main text .
the histogram curves are normalized in order to enclose the same integral area .
gaussian fits are added as guides to the eye .
obviously , the height histogram of graphene is broader exhibiting a full width at half maximum ( fwhm ) of 0.78 nm , while sio@xmath0 exhibits a fwhm of only 0.48 nm . while the latter value is quite robust across the surface , the former value is changing by about 25 % across the surface , but is on average larger than the fwhm on sio@xmath0 .
this demonstrates that the rippling of graphene is not limited to substrate corrugations .
( color online ) height histograms of sio@xmath0 ( blue ) and graphene ( red ) corresponding to the images in fig . 3 of the main text .
on average , the graphene histogram exhibits a larger fwhm value compared to sio@xmath0 . ]
figure [ figs2 ] shows two afm images of the sio@xmath0 substrate obtained in tapping mode by two different tips .
representative scanning electron microscopy ( sem ) images of the tips provided by the supplier are added as insets .
they are cross - checked by sem images after the measurements .
these images are in agreement with the supplier images , but due to a moderate background pressure , the ultrasharp w - tip fastly burnt away during sem imaging . the left afm image is measured by a cantilever with a si tip having an apex radius of about 10 nm , while the right afm image is measured with a w tip of nominal tip apex radius of 1 nm mounted at the bottom of the si tip .
both images exhibit a very similar lateral scale of the corrugation .
the rms values are 220 pm and 270 pm , respectively , showing that the tip properties do not significantly change the observed corrugation of the substrate .
we cross - checked this result using more than ten images obtained with both tips , which lead to fluctuations in rms values of less than @xmath25 10% .
figure [ figs3 ] shows a comparison of the correlation functions of the topographic images of graphene ( top row ) and sio@xmath0 ( bottom row ) .
the left column shows the correlation functions directly obtained from the topography .
the middle and right column show the high - frequency and low - frequency part of the correlation function separated by a smooth first - order butterworth filter with cutoff wavelength @xmath20 nm . while the large wavelengths visible in the right columns are quite similar in both images , the rather regular short wavelength pattern ( middle column ) is only observed on graphene .
note that correlation functions within the same columns are scaled to the same contrast , i. e. , their brightness is directly comparable .
figure [ figs4 ] shows three images ( top row ) obtained in different areas of the monolayer graphene , separated by several @xmath9 m .
they exhibit different rms values of 270 pm , 220 pm and 360 pm ( left to right ) , respectively .
the middle row shows the corresponding correlation functions .
the bottom row shows the same correlation functions after high - pass filtering using the identical procedure as in figure 3 .
notice that the scales of the images are different .
while the long - range corrugation ( middle row ) is oriented in different directions , the short range corrugations ( bottom row ) exhibit the same wave length of about 15 nm in all three areas and show a similar orientation as marked by the white lines .
the latter indicates that the the short - scale corrugation is related to the atomic directions in graphene .
but since the directions of the rippling are fluctuating slightly across the sample similar to the orientations of molecules in liquid crystals , more data are required in order to substantiate the relation between atomic directions and preferred directions of rippling .
figure [ figs5 ] shows topographic images of different areas of the sio@xmath0 substrate ( top row ) and the corresponding correlation functions ( bottom row ) .
they exhibit similar length scales of corrugation but clearly different orientations , which might be due to the polishing procedure of the sample .
figure [ figs6 ] shows the radially averaged fourrier transforms ( ft ) of the topographic images of graphene ( black curve ) and sio@xmath0 ( grey curve ) .
the ft curves are based on images of the same lateral size and resolution .
the curves are averaged using 10 images of 200 @xmath24 200 nm@xmath7 from different areas of the surfaces . obviously , the corrugation of sio@xmath0 is larger at large wave length ( small wave number ) down to 30 nm by up to 66 % , while the corrugation on the graphene is larger at smaller wavelengths ( larger wave numbers ) by up to a factor of 2.6 .
the individual fts of different sio@xmath0 areas are nearly indistinguishable , i.e they are not deviating from the width of the grey curve in fig .
5 . however , the fts of different graphene areas are shifted vertically due to the fluctuating rms value of corrugation , which leads to crossing points of the two curves varying between wavelengths of 20 nm and 40 nm .
importantly , all of the graphene curves show one crossing point with the averaged sio@xmath0 curve . in order to observe the influence of the graphene film morphology on raman scattering , we compared raman spectra ( laser wave length 532.1 nm ) , taken in identical experimental conditions , of a monolayer flake which is partly suspended and one that follows the substrate ( fig .
[ figs7 ] ) .
the surface structure of these samples was determined using stm measurements ( insets fig .
[ figs7]a ) .
the raman spectra were recorded before and after the stm measurements giving the same result .
a difference between both spectra regarding the positions of the 2d and g peaks is clearly visible .
the wave number of the 2d peak is 2686.7 @xmath26 for the partly suspended graphene flake and 2670.5 @xmath26 for the flake following the substrate .
hence , it results that the change in morphology causes a shift of the 2d peak by 16.2 @xmath26 ( fig .
[ figs7]a ) .
a similar effect was observed at the g peak which is shifted by 5.5 @xmath26 to lower values for the flake following the substrate ( fig .
[ figs7]b ) .
the d peak ( about 1350 @xmath26 ) is barely visible for both samples indicating a good quality of our graphene .
raman spectra of several of our graphene samples ( about 15 ) prepared in nominally identical conditions show more frequently a non shifted 2d line with an average value of [email protected] @xmath26 .
hence , we conclude that our samples mostly exhibit intrinsic rippling .
if we compare with wave numbers of the 2d line within the literature using only the data taken with a similar laser wave length ( 514 - 532 nm ) @xcite , we find 2680 - 2700 @xmath26 .
hence , we believe that the majority of samples investigated so far exhibit intrinsic rippling as displayed in the right inset of figure 7a . | using scanning tunneling microscopy ( stm ) in ultra high vacuum and atomic force microscopy , we investigate the corrugation of graphene flakes deposited by exfoliation on a si / sio@xmath0 ( 300 nm ) surface . while the corrugation on sio@xmath0 is long - range with a correlation length of about 25 nm , some of the graphene monolayers exhibit an additional corrugation with a preferential wave length of about 15 nm .
a detailed analysis shows that the long range corrugation of the substrate is also visible on graphene , but with a reduced amplitude , leading to the conclusion that the graphene is partly freely suspended between hills of the substrate .
thus , the intrinsic rippling observed previously on artificially suspended graphene can exist as well , if graphene is deposited on sio@xmath0 .
since it was believed , based on the mermin - wagner theorem , that two - dimensional ( 2d ) crystals are not stable at finite temperature @xcite , it came as a surprise that monolayer graphene could be stabilized on a si / sio@xmath0 substrate @xcite .
due to its peculiar properties like , e.g. , a linear dispersion leading to klein tunneling @xcite , high room - temperature mobility allowing quantum hall steps at 300 k @xcite , a low spin - orbit interaction beneficial for spintronic devices @xcite or tunable spin - polarized edge states @xcite , graphene studies have become a major issue in solid - state physics .
already the first transport results not exhibiting weak localization lead to the speculation of a curved surface @xcite acting as a phase - breaking field @xcite .
such a curvature was indeed observed by microscopic electron diffraction of suspended monolayer graphene sheets @xcite .
the lateral wavelength of the isotropic curvature is estimated to be @xmath1 nm with an amplitude of @xmath2 nm .
the rippling has been reproduced theoretically by monte - carlo simulations with a preferential wavelength of 8 nm barely depending on temperature @xcite .
it is argued that the anharmonic coupling between bending and stretching modes in graphene causes the rippling and is responsible for the stability of the 2d crystal .
moreover , it has been shown experimentally that rippling can ultimately limit the mobility of graphene at 300 k , if defects are avoided @xcite however , the corrugation properties of graphene deposited on a substrate as typically used in transport experiments are not clarified .
in particular , previous scanning probe studies @xcite revealed only corrugations , which are attributed to the roughness of the underlying substrate , except for a torn flake manipulated by afm @xcite .
thus , it is unclear , if the intrinsic tendency for rippling persists on the substrate , thereby ultimately limiting mobility and influencing weak localization , which has meanwhile been found , at least , for some samples @xcite . here
, we demonstrate that a regular short wave - length corrugation with @xmath3 nm and @xmath2 nm , which is not induced by the substrate , can even prevail the substrate corrugation .
since @xmath4 and @xmath5 are close to the values found on suspended graphene @xcite , we attribute this additional corrugation to the intrinsic rippling of graphene @xcite .
moreover , we find that the long - range corrugation implied by the substrate has lower amplitude on graphene than on sio@xmath0 suggesting that our high - mobility graphene is partly suspended between hills of the substrate , which might favor the development of the intrinsic short - scale rippling .
the graphene sample is fabricated by the mechanical exfoliation technique as described in @xcite .
using an optical microscope , a graphene flake containing a monolayer region is identified .
raman spectroscopy is used to confirm the number of layers with the help of the 2d line @xcite .
fig .
[ fig1](a ) shows three spectra measured in different areas of the flake visible in fig .
[ fig1](b ) .
in addition to a large monolayer area , we find smaller bi- and multilayer areas as marked by 1l , 2l and mul , respectively .
next , gold contacts with a 10 nm cr seed layer were deposited and structured with a lift - off process .
except for one side , the graphene sample is completely surrounded by the gold electrode as shown in fig .
[ fig1](b ) .
the mobility of graphene samples prepared identically but with several contacts is @xmath6 m@xmath7/(vs ) at 300 k @xcite , i.e. comparable with high - mobility values obtained by other groups @xcite . in order to remove residual resist and adsorbates ,
the sample was rinsed in isopropanol and acetone , baked out to 150@xmath8 c in air for four hours and , additionally , baked out to the same temperature in ultrahigh - vacuum ( uhv ) for three hours .
the gold film served as the electrical contact for scanning tunneling microscopy ( stm ) and was used to prepare the stm tip ( etched w ) by voltage pulses . in order to find the graphene within the uhv - stm
, we used an optical microscope with a focal length of 30 cm and 5 @xmath9 m lateral resolution .
fig .
[ fig1](c ) shows a microscopic image of the tip approaching the graphene flake .
the home - built stm features an @xmath10 stage for lateral positioning and operates in uhv ( @xmath11 pa ) as described elsewhere @xcite .
the @xmath10 stage has been checked to move accurately within 10 % .
topographic images are recorded applying a bias @xmath12 to the tip .
spectroscopic @xmath13 curves are measured by lock - in technique using a modulation voltage @xmath14 after stabilizing the tip at voltage @xmath15 and current @xmath16 .
stm images obtained at the intentionally irregular edge of the gold contact are displayed in the insets of fig .
[ fig1](b ) .
they are used for orientation by stm and aid in finding monolayer , bilayer and multilayer areas on the flake .
( color online ) ( a ) raman spectra using laser light of wavelength 532.1 nm .
the spectra are recorded on different areas of the graphene flake and the attribution to monolayer , bilayer and multilayer is marked .
( b ) optical image of the graphene sample with gold contacts .
the different areas of monolayer ( 1l ) , bilayer ( 2l ) and multilayer ( mul ) as identified by raman spectroscopy are indicated .
insets show two stm images acquired at the edge of the gold contact , which are used for orientation .
( c ) microscopic image of the stm tip in tunneling contact with the graphene in uhv recorded with a long - range optical microscope . ]
atomic force microscopy images are taken under ambient conditions in tapping mode using either a commercial cantilever with a silicon tip or an ultrasharp tungsten tip with a tip radius of 1 nm , which is attached at the bottom of a si cantilever @xcite .
imaging has been carried out in the attractive regime @xcite with an oscillation frequency slightly above resonance .
fig .
[ fig2](b ) and ( c ) show two atomically resolved stm images recorded on the monolayer ( 1l ) and the multilayer ( mul ) of graphene , respectively .
one observes a hexagonal pattern on the monolayer and a triangular pattern on the multilayer in accordance with previous results @xcite .
the atomic resolution is still perceptible at lower resolution in fig .
[ fig2](a ) and ( d ) , but an additional irregular corrugation appears .
line sections shown in fig .
[ fig2](e ) reveal that the corrugation height on this monolayer region is about 0.6 nm , a factor of three larger than on the multilayer region .
notice that corrugation heights pretended by charged defects are typically smaller by more than an order of magnitude @xcite . a representative @xmath17 curve of the monolayer , which is known to represent the local density of states ( ldos ) @xcite , is shown in fig . [ fig2](f ) .
as expected for graphene , it shows a v - like shape with the dirac point at about @xmath18 mv .
it does not change significantly across the area depicted in fig .
[ fig2](d ) .
note that graphene on sic(0001 ) exhibits a more complicated @xmath13 spectrum @xcite . note also , that we did not observe the phonon gap found at 4 k recently @xcite , although we used several different micro- and macrotips .
this is tentatively attributed to the different temperature of the two experiments .
the good representation of the graphene ldos by @xmath17 and the continuous atomic resolution across fig .
[ fig2](d ) demonstrates the absence of resist on this part of the surface .
indeed , we found several large areas without resist , but still also areas where a remaining resist is apparent within the stm images .
the fact that the dirac point is observed close to 0 v shows a negligible influence of charging adsorbates , which are probably removed in uhv .
fig .
[ fig3](a ) shows a large area stm image of the graphene monolayer .
one observes a rather regular corrugation with amplitudes of more than 1 nm and a preferential distance between hills of about 15 nm .
the histogram @xcite is gaussian with a full width at half maximum ( fwhm ) of 0.78 nm .
the rms roughness is determined to be 0.36 nm .
the latter two values fluctuate across the monolayer area by about 25 % ( average rms roughness 0.32 nm ) , but the corrugation length scale remains constant . for comparison , fig .
[ fig3](c ) shows an afm image of the bare sio@xmath0 surface recorded in tapping - mode .
it is obvious that the corrugation on the substrate is lower and exhibits a larger length scale than the corrugation on graphene .
the fwhm of the histogram of fig .
[ fig3](c ) is 0.48 nm and the rms roughness is 0.22 nm ( average of all data @xmath19 nm ) .
we used four different tips , including two ultrasharp tips with a tip radius of 1 nm @xcite , but always observed the same corrugation height and length scale , which , in addition , is in good agreement with previous results @xcite .
careful inspection of the graphene monolayer in fig .
[ fig3](a ) and ( b ) shows a modulation on top of the short - range corrugation exhibiting a similar length scale as the corrugation on the sio@xmath0 substrate . in order to disentangle the two contributions , we used two - dimensional autocorrelation functions of the images ( a ) and ( c ) @xcite , which are shown in fig .
[ fig4](a ) and ( b ) .
note that both figures use the same color scale , i. e. the intensities are directly comparable . a long - range structure represented by a central spot surrounded by four bright areas is visible in both images although with weaker intensity for graphene ( a ) .
fig .
[ fig4](c ) shows the autocorrelation function of the graphene after removing the long - range part by high - pass filtering using a smooth ( first order butterworth ) cutoff at wavelength @xmath20 nm .
a preferential short - range distance is clearly visible by the strong spots around the center , but there is additional multiple correlation albeit not with high symmetry .
notice that creep and drift effects are carefully removed from the stm - image using the atomic resolution .
we checked the preferential directions visible in both correlation patterns .
they are rotating arbitrarily across the sio@xmath0 , but we find indication for a preferential orientation on graphene @xcite .
fig . [ fig4](d ) shows radial line sections averaged over all angles of the correlation functions .
the sio@xmath0 ( black curve ) exhibits a correlation length of about 25 nm comparable to @xcite and , in addition , a very weak maximum at about 50 nm indicating a slight preferential distance between hills .
the gray curve shows the radial line section of the high - pass filtered image of graphene visible in fig .
[ fig4](c ) .
it exhibits a damped oscillation with a wave length of about 14 nm looking very similar to correlation functions of liquids . for comparison ,
the radial line section of the autocorrelation function of the afm image after applying the same high - pass filter is shown as a dotted line and does not exhibit any structure .
we observe the damped oscillation also after high - pass filtering the original stm image in fig .
[ fig3](a ) and determining the autocorrelation function afterwards .
in addition , we checked that the obtained wave length is quite robust with respect to slight parameter changes of the filtering process .
depending on the filtering procedure and the selection of the minima or maxima used for determining the wave length , @xmath4 is varying by at maximum @xmath21 nm with a mean value of @xmath22 nm .
( color online ) ( a ) 2d autocorrelation function of the drift corrected stm image of monolayer graphene as shown in fig .
[ fig3](a ) .
( b ) 2d autocorrelation function of the afm image of the sio@xmath0 surface as shown in fig .
[ fig3](c ) , same color scale as in ( a ) .
( c ) fft high pass filtered image ( 1 .
order butterworth , 20 nm ) of the 2d autocorrelation function shown in ( a ) .
( d ) radial line sections of the unfiltered afm autocorrelation image ( b ) ( black ) , of the high pass filtered stm autocorrelation image ( c ) ( gray ) , and of the high pass filtered afm autocorrelation image ( dotted ) .
the inset shows a larger view of the area marked by the dashed line .
the black and gray graphs are offset vertically for clarity . ] from these results , we conclude that monolayer graphene can exhibit a rather regular , liquid - like short - scale corrugation not induced by the substrate .
it is similar in height and wavelength to the one observed on suspended graphene @xcite and calculated for freely suspended graphene by atomic monte - carlo methods @xcite .
this observation suggests that the graphene might be partly suspended even on the sio@xmath0 . in order to verify this , we compared the fourier components of the images of graphene and sio@xmath0 within the large wavelengths region @xmath23 nm @xcite .
we indeed find that graphene exhibits an up to 40 % lower amplitude at all wavelengths within this wavelength range .
this implies that the graphene does not follow the substrate corrugations exactly , but instead includes areas not in contact with the substrate .
we believe that this partly free - standing configuration is the prerequisite for the intrinsic rippling of graphene .
indeed , we have also found other graphene flakes prepared nominally identically by the same person which did not show the intrinsic rippling .
this highlights that tiny details of the preparation process can have large impacts on the physical properties , maybe explaining the contradicting weak localization properties in different studies @xcite .
a comparison of raman data of several samples indicates that samples with intrinsic rippling are more frequent than samples without @xcite .
future systematic studies using our method to determine the intrinsic rippling are required . in summary
, we have investigated high - mobility graphene flakes on sio@xmath0 by uhv - stm and scanning tunneling spectroscopy .
a mesoscopic corrugation decreasing in amplitude with increasing thickness not induced by the substrate is identified .
it is rather regular exhibiting liquid - like correlation properties with a preferential wavelength of about 15 nm and a rms roughness of 0.32 nm . since the long - range corrugation of the substrate is additionally visible on graphene , but with smaller amplitude than on the substrate
, we conclude that the rippled graphene is partly freely suspended .
we gratefully acknowledge useful discussions with g. gntherodt , u. d. schwarz and h. kurz and financial support by for 912 , project tp 6 of the deutsche forschungsgemeinschaft and by the german federal ministry of education and research ( bmbf ) under contract number nknf 03x5508 ( `` alegra '' ) . |
deuteron - gold collisions at rhic provide a means to explore nuclear effects on the initial - state parton densities in the nucleus , which is vitally important for understanding the baseline production in heavy - ion collisions .
rhic experiments have shown that single inclusive hadron yields in the forward ( deuteron ) rapidity direction for @xmath7 gev @xmath0aucollisions are suppressed relative to @xmath6+@xmath6collisions @xcite .
the mechanism for the suppression has not been firmly established .
many effects have been proposed for this suppression , such as gluon saturation @xcite , initial state energy loss @xcite , parton recombination @xcite , multi - parton interactions @xcite , and leading and higher - twist shadowing @xcite .
one set of measurements that might help to distinguish between the competing models is forward azimuthally correlated di - hadron correlation functions , which directly probe di - jet production through their 2@xmath82 back - to - back peak at @xmath9 .
this technique has been used extensively at rhic and is described in detail elsewhere @xcite .
the di - hadron results presented here were obtained from @xmath6+@xmath6and @xmath0auruns in 2008 with the phenix detector and include a new electromagnetic calorimeter , the muon piston calorimeter ( mpc ) , with an acceptance of @xmath10 in pseudorapidity and @xmath11 .
di - hadron measurements can probe more precise ranges of parton @xmath3 in a gold nucleus than do single hadron probes ( e.g. , ) . at forward rapidities
, a single hadron probe will cover a very broad range of x , @xmath12 , thus mixing together the shadowing , anti - shadowing , and even emc effects @xcite .
azimuthally correlated di - hadron measurements also enhance the di - jet fraction in the event selection , since one selects only the back - to - back hadrons .
+ @xmath6(open points ) , @xmath0auperipheral ( 60 - 88% , triangles ) , and @xmath0aucentral collisions ( 0 - 20% , solid points ) at @xmath13 gev .
the trigger @xmath2 ranges from 1.1 to 7 gev / c and the associated s have @xmath14 gev / c .
systematic errors of up to 30% on the near side ( @xmath15 ) have not been shown .
the subtracted pedestals , b@xmath16 , are shown for each case.,title="fig : " ] + + @xmath6(open points ) , @xmath0auperipheral ( 60 - 88% , triangles ) , and @xmath0aucentral collisions ( 0 - 20% , solid points ) at @xmath13 gev .
the trigger @xmath2 ranges from 1.1 to 7 gev / c and the associated s have @xmath14 gev / c .
systematic errors of up to 30% on the near side ( @xmath15 ) have not been shown .
the subtracted pedestals , b@xmath16 , are shown for each case.,title="fig : " ] by performing several correlation measurements with particles at different and rapidities , one can systematically scan different @xmath3 ranges with an observable that is enhanced for the leading - order perturbative qcd component . probing
the @xmath3 dependence of the effect is an important test since most models predict that any effects should be stronger at smaller @xmath3 .
particles at higher pseudorapidities are produced from smaller @xmath3 , so measuring hadrons from more forward rapidities should probe smaller x.
for this analysis , back - to - back -or hadron - pairs are measured with one particle at mid - rapidity , and the other at forward rapidity .
back - to - back cluster - pairs are also measured where both are in the forward rapidity region .
the clusters are reconstructed from the energy deposit of photons in the mpc , and are estimated to be at least 80% dominated by s , with the remainder coming from single photons from decays of @xmath17 s and from direct photons .
further details of the analysis are available in @xcite .
as shown in figure [ fig : mpc_corr ] , the away - side peak for @xmath0aucentral collisions appears significantly suppressed compared to @xmath6+@xmath6collisions and peripheral @xmath0aucollisions .
this effect is large for the mid - forward di - hadron correlations and becomes even larger for the forward - forward correlations . within large errors ,
the gaussian widths of the away - side correlation peak for the mid - forward di - hadron correlations remain the same between @xmath6+@xmath6and central @xmath0au . for the forward - forward case , uncertainties in the pedestal level from the underlying event and the strong suppression of the away - side peak make extracting the width unreliable . for this case
, the away side peak width in central @xmath0aucollisions is allowed to vary up to twice as much as in @xmath6+@xmath6when accounting for this systematic uncertainty . versus @xmath18 for peripheral ( 60 - 88% , in red ) and central ( 0 - 20% , in blue ) @xmath0aucollisions at @xmath13 gev , and compared to the eps09 lo @xmath19 curves at a scale @xmath20 = 4 gev@xmath21 @xcite . ]
the observed suppression is quantified by studying the relative yield , @xmath22 @xcite , of correlated back - to - back hadron pairs in @xmath0au collision compared to @xmath6+@xmath6collisions scaled with the average number of binary nucleon collisions @xmath23 , @xmath24 this is simply the analog of the usual nuclear modification factor but for hadron pairs .
the @xmath25 are the @xmath6+@xmath6or @xmath0auinelastic cross - sections , while @xmath26 is the cross - section for di - hadron pair production .
the is calculated using the correlated away side peak after subtracting the pedestal @xmath27 . decreases with increasing number of binary collisions , @xmath23 , or equivalently with increasing nuclear thickness .
the suppression also increases with decreasing particle @xmath2 and is significantly larger for forward - forward hadron pairs than for mid - forward pairs . the observed suppression of @xmath22 versus nuclear thickness , @xmath2 and @xmath17 points to large cold nuclear matter effects arising at low parton momentum fractions x in the nucleus probed by the deuteron .
this trend is seen more clearly in fig .
[ fig : jdax ] where @xmath22 is plotted versus @xmath28 for all pair selections in @xmath17 and @xmath2 . in the case of 2@xmath82 leading order ( lo ) processes , the variable @xmath18 is lower than @xmath29 by the mean fragmentation fraction , @xmath30 , of the struck parton in the au nucleus . since @xmath18 is an entirely experimental defined quantity , it should be reproducable in any theoretical framework .
in a leading order pqcd picture , the variable is @xmath31 for partons a+b going to outgoing jets c+d , which then fragment to hadrons with longitudinal fractions @xmath32 , @xmath33 . in the above convolutions over p+p and d+a ,
most of the terms are expected to be roughly similar between p+p and d+au except for the nuclear gluon pdf .
naively , might be largely dominated by the modification to the nuclear gluon parton distribution function ( pdf s ) , since most of the events with di - hadrons at forward rapidities consist of a high - x parton from the deuteron and a low - x gluon from the gold nucleus . assuming this to be true
then @xmath34 in figure [ fig : jdax ] the values are overlaid with the eps09 curves @xcite .
the values for the peripheral bins are above the best fit eps09 , while the central bins are below .
the eps09 curves are taken largely from nuclear deep inelastic scattering and represent an averaged value of over all centralities .
the values for the most central bin at the lowest x are well below the eps09 curves .
this is qualititatively consistent with the expectations for the color - glass condensate @xcite , which posits an extreme form of shadowing due to the onset of gluon saturation .
if nature is kind and this data can be interpreted in terms of a simple lo pqcd picture , it may be possible to extract , which is extremely important for understanding the quark gluon plasma since it forms the baseline for production in heavy ion collisions .
in addition , if the large suppression of observed in central @xmath0aucollisions is from gluon saturation , it may be possible to study the dependence of that saturation on the thickness of the nucleus .
one possible test of whether these ideas are correct would be to use extractions of from this data to predict @xmath35 data in @xmath0aucollisions from phenix .
20 i. arsene _ et al._[brahms collaboration ] , phys . rev .
* 93 * ( 2004 ) 242303 j. adams _ et al._[star collaboration ] , phys .
* 97 * ( 2006 ) 152302 s. adler
_ et al._[phenix collaboration ] , phys .
* 94 * ( 2005 ) 082302 l. d. mclerran and r. venugopalan , phys .
d * 49 * ( 1994 ) 3352 d. kharzeev , e. levin and l. mclerran , nucl .
a * 748 * ( 2005 ) 627 i. vitev phys .
c * 75 * ( 2007 ) 064906 l. frankfurt and m. strikman , phys .
b * 645 * ( 2007 ) 412 r. c. hwa , c. b. yang and r. j. fries , phys .
c * 71 * ( 2005 ) 024902 m. strikman and w. vogelsang , phys .
d * 83 * , 034029 ( 2011 ) v. guzey , m. strikman and w. vogelsang , phys .
b * 603 * , 173 ( 2004 ) j. w. qiu and i. vitev , phys .
b * 632 * ( 2006 ) 507 s. adler _ et al._[phenix collaboration ] , phys .
c. * 73 * ( 2006 ) 054903 s. adler _ et al._[phenix collaboration ] , phys . rev .
lett . * 96 * ( 2006 ) 222301 a. adare _ et al . _
[ phenix collaboration ] , submitted to phys .
_ preprint _ nucl - ex:1105.5112 a. adare _ et al._[phenix collaboration ] , phys . rev . c. * 78 * ( 2008 ) 014901 k. j. eskola , h. paukkunen and c. a. salgado , jhep * 0904 * ( 2009 ) 065 [ arxiv:0902.4154 [ hep - ph ] ] | we present measurements by the phenix experiment at rhic of di - hadron pair production in @xmath0aucollisions where the particles in the pair are varied across a wide range of pseudorapidity , out to @xmath1 . with di - hadrons
, varying the @xmath2 and rapidity of the particles in the di - hadron pair allows studying any effects as a function of partonic @xmath3 in the nucleus .
these di - hadron measurements might probe down to parton momentum fractions x @xmath4 @xmath5 in the gold nucleus , where the interesting possibility of observing gluon saturation effects at rhic is the greatest .
our measurements show that the correlated yield of back - to - back pairs in @xmath0aucollisions is suppressed by up to an order of magnitude relative to @xmath6+@xmath6collisions , and increases with greater nuclear path thickness and with a selection for lower x in the au nucleus . |
the theoretical prediction and subsequent experimental verification of the conducting surface state of a strong topological insulator ( ti ) continues to generate a lot of excitement in physics ; for reviews , see refs . .
in a ti , strong spin - orbit couplings and band inversion conspire to produce a time - reversal invariant topological state different from a conventional band insulator . using bi@xmath0se@xmath1 as a weakly correlated reference ti material with rather large bulk gap @xmath2 mev ,
surface probe experiments have provided strong evidence for the topologically protected gapless surface state.@xcite the measured spin texture of the surface state is consistent with predictions obtained for two - dimensional ( 2d ) massless dirac fermions . under this `` relativistic '' description ,
the spin direction is locked to the surface , and the surface state is stable against the effects of weak disorder and weak interactions ( topological protection).@xcite it is thus useful to first study the simplest case of a noninteracting disorder - free model , which is the case investigated below . because of the residual bulk conductivity of the presently available ( nominally insulating ) ti samples , it has been difficult to experimentally extract the surface contribution to the electrical conductivity .
one attempt to improve the situation is to consider mesoscopic samples , where the surface - to - volume ratio is more advantageous . in particular , thin - film geometries@xcite and quasi-1d nanowires ( `` ribbons'')@xcite have been studied experimentally .
signatures for aharonov - bohm interference effects associated with the topological surface state in bi@xmath0se@xmath1 nanowires were reported,@xcite cf .
also related experiments for sb@xmath0te@xmath1 nanowires.@xcite for infinitely long and circulary symmetric topological nanowires , band structure calculations predict a multi - channel waveguide where all surface modes are gapped because of spin - surface locking.@xcite with @xmath3 denoting the conserved momentum along the wire axis ( taken along the @xmath4 direction ) , and @xmath5 the half - integer total angular momentum , the dispersion relation of these modes is @xmath6 where @xmath7 throughout and @xmath8 for conduction and valence band , respectively .
the fermi velocities @xmath9 and @xmath10 differ because the bulk dispersion relation is anisotropic , see below , and the nanowire radius is @xmath11 .
note that there is a minimal gap @xmath12 for the surface modes since @xmath5 is half - integer . for reasonable values of @xmath11
, we have @xmath13 .
experiments probing quantum dot physics in finite - length ti nanowires are expected to yield new insights into the exciting physics of tis , in close analogy to semiconductor nanowires and carbon nanotubes where such experiments have been highly successful.@xcite to prepare the ground , we here consider the band structure for ti quantum dots with a finite - length nanowire geometry . for a spherical ti
dot , the band structure was worked out before.@xcite however , we draw attention to several features that only arise when the surface contains sharp edges , i.e. , non - differentiable parts , as is the case for the cylinder .
we employ three different and independent approaches to understand ti quantum dot energy levels and their spin texture : ( 1 ) for a cylindrical ti nanowire of length @xmath14 and radius @xmath11 closed by flat caps , we have performed detailed numerical calculations for the energy spectrum and the spin texture of the eigenstates based on the effective low - energy theory of zhang _
et al._;@xcite material parameters were chosen for bi@xmath0se@xmath1 as quoted in ref . .
( 2 ) an analytical approach starting from a surface dirac fermion description has been developed for the same geometry .
most of our numerical results can thereby be quantitatively reproduced within an analytical theory .
( 3 ) we have also studied a microscopic tight - binding model for a strong ti in the finite - size nanowire geometry and find qualitatively similar results .
the main conclusions reached from these three approaches are as follows : first , longitudinal momentum quantization implies a discrete sequence of energy levels which can ( roughly ) be approximated by letting @xmath15 ( @xmath16 ) in eq . .
remarkably , in addition we find unconventional `` zero - momentum '' states ( where , formally , @xmath17 ) . in such a state , the charge and spin densities are almost homogeneous along the @xmath4-direction .
furthermore , there are subgap states energetically located within the surface mode gap ( @xmath18 ) .
these states are localized near both caps and show interesting spin texture .
second , we observe significant out - of - surface components for the spin density associated with all energy eigenstates .
note that an out - of - plane spin texture is only expected when trigonal warping effects are important,@xcite see also very recent experimental results reporting such features.@xcite however , to lowest order in momentum ( around the @xmath19 point ) , trigonal warping can be neglected , while we persistently find broken spin - surface locking also at the lowest energy scales .
our observations are instead related to the presence of non - differentiable sections of the surface , where the wavefunctions in the `` trunk '' and `` cap '' regions of the cylinder have to be matched .
such non - differentiable surface parts also appear in the samples studied in refs . , and therefore surface probe experiments for these devices
could directly test our prediction of broken spin - surface locking .
for differentiable closed surfaces , we expect spin - surface locking to stay intact .
in fact , the explicit solution of the problem in a spherical geometry exhibits spin - surface locking.@xcite we note in passing that for a flat ti surface , a time - dependent out - of - plane spin component can also be generated by elastic disorder .
however , this component will precess around the momentum - dependent spin - orbit axis ( which lies in the plane ) and averages to zero on time scales corresponding to the inverse fermi energy.@xcite the structure of the remainder of this paper is as follows . in sec .
[ sec2 ] , we describe the results of our numerical calculations for cylindrical ti nanowire dots based on the effective low - energy theory . in sec .
[ sec3 ] , an analytical approach based on the surface dirac fermion picture is described for the same geometry .
the spectrum and the spin density profile for the resulting eigenstates will be derived , and the results are compared to the numerical findings in sec .
[ sec2 ] . in sec .
[ sec4 ] , we turn to the tight - binding calculation and compare those computations to the previous results .
finally , we conclude in sec .
appendix [ appa ] contains a derivation of the surface dirac fermion hamiltonian for an infinitely long nanowire using the approach of ref . .
in this section , we compute the band structure of a cylindrical nanowire of length @xmath14 and radius @xmath11 from the effective low - energy theory of zhang _ _ et al.__@xcite using parameters for bi@xmath0se@xmath1 .
up to terms of order @xmath20 with @xmath21 , the low - energy bulk hamiltonian employing the four bands energetically closest to the @xmath19 point is @xcite @xmath22\tau_x,\ ] ] where @xmath23 and @xmath24 , with @xmath25 .
this also defines the fermi velocities @xmath26 ( in @xmath4 direction ) and @xmath27 ( in the @xmath28 plane ) , see eq . .
the basis states encoding the spin - parity structure , where pauli matrices @xmath29 ( @xmath30 ) act in spin ( parity ) space and @xmath31 ( @xmath32 ) denotes the respective unit matrix , are explicitly given in ref . .
this work also describes the extension of eq . to the case of eight bands or towards including trigonal warping . using the parameters in ref . ,
listed for convenience also in table [ tab0 ] , the criterion for a ti phase@xcite is satisfied .
one therefore must have an odd number of conducting surface modes when boundaries are present . here
we analytically derive eq .
( [ hd ] ) for an infinite nanowire using the gap - inversion model of ref . .
our starting point is eq . to linear order in @xmath33 ( we put @xmath34 here ) , @xmath35 \tau_x
.\ ] ] following ref .
, we assume that the gap parameter changes sign at @xmath36 , i.e. , @xmath37 with @xmath38 . for @xmath39 ( @xmath40 ) , the material is then in the topologically nontrivial ( trivial ) phase . in the 1d representation , for given @xmath5 and @xmath3 , solutions to the dirac equation take the form @xmath41 , with the 1d radial dirac equation @xmath42 here @xmath43 corresponds to @xmath44 \tau_x .\ ] ] for @xmath43 , there are two solutions @xmath45 where the first ( second ) spinor refers to parity ( spin ) space , @xmath46 , @xmath47 , and @xmath48 for @xmath40 , we have to replace @xmath49 and @xmath50 in the spin part , where @xmath51 and @xmath52 are modified bessel functions .
the general solution is then given by @xmath53 where the coefficients @xmath54 and @xmath55 are obtained by requiring continuity of the wave function at @xmath36 .
this results in a linear system of equations for the coefficients , which has a nontrivial solution under the condition @xmath56 where all bessel functions have the argument @xmath57 and @xmath58 is evaluated for @xmath59 .
assuming @xmath60 and using the asymptotic form of the bessel functions , we find @xmath61 and thus @xmath62 .
the corresponding @xmath43 wavefunctions , @xmath63 , are given by eq .
with @xmath64 , @xmath65 , and @xmath66 for small @xmath67 , the effective surface hamiltonian , eq .
( [ hd ] ) , is obtained by projecting eq . onto the subspace spanned by the above @xmath43 states .
we thereby find @xmath68 where the @xmath69 are pauli matrices in the zero - momentum subspace . in this way
, all combinations of spin - parity matrices , @xmath70 , can be represented in the truncated basis .
for @xmath60 , we obtain the results in table [ tab1 ] .
of course , this representation is not single - valued , i.e. , there is no one - to - one correspondence between the @xmath71 and the @xmath72 matrices . in particular , the spin operators @xmath73 and the parity operator @xmath74 are entangled , since @xmath75 and @xmath76 . taking @xmath77 and @xmath78 , cf . the last column of table [ tab1 ] , one gets the surface hamiltonian quoted in the main text , see eq . .
because this representation is multi - valued , the replacement of @xmath79 by products of spin and parity matrices causes a double - counting of states if used naively .
this double - counting problem is , however , not severe and can be circumvented , as we discuss in sec .
[ sec3c ] in the main text .
we note that a theory for quantum dots based on the 2d semiconductors hgte and hgs ( which may have topological phases ) was recently put forward by n. malkova and g.w .
bryant , phys .
b * 82 * , 155314 ( 2010 ) . | we consider the energy spectrum and the spin - parity structure of the eigenstates for a quantum dot made of a strong topological insulator .
using the effective low - energy theory in a finite - length cylinder geometry , numerical calculations show that even at the lowest energy scales , the spin direction in a topologically protected surface mode is not locked to the surface .
we find `` zero - momentum '' modes , and subgap states localized near the `` caps '' of the dot .
both the energy spectrum and the spin texture of the eigenstates are basically reproduced from an analytical surface dirac fermion description .
our results are compared to microscopic calculations using a tight - binding model for a strong topological insulator in a finite - length nanowire geometry . |
the process of phase - separation occurring in binary systems quenched below the coexistence line has been matter of intensive investigations since many years .
the overall phenomenology is nowadays reasonably well understood . in an early stage ,
whose properties depend on the system specific details , domains of the two phases form ; the relative concentration of the two species well inside the domains quickly saturates to a value very similar to the equilibrium composition .
then a crossover leads to the late stage dynamics which is characterized by the growth of ordered regions of typical size @xmath4 and by a statistically invariant morphology .
this phenomenology is at the basis of the well known scaling property according to which a single length @xmath4 is asymptotically relevant : system configurations at different times are simply related by a scale transformation with a scale factor @xmath4 . in particular , for the equal time density - density correlation function one has g(r , t)=g , [ sca ] meaning precisely that @xmath5 is invariant as a function of scaled lengths @xmath6 .
the growth of @xmath4 is of the power law type @xmath7 .
the dynamic exponent @xmath8 depends on the specific microscopic mechanisms responsible for the segregation process ; in general different dynamical regimes can be observed , characterized by different values of @xmath8 , by changing the experimental time window @xcite . when the system enters the late stage the memory of the initial state
is lost and , due to the presence of one single growing length , its properties become universal .
then the dynamical exponents , the scaling function @xmath9 and other observables become independent on the system specific details and are only determined by a few relevant parameters . besides the spatial dimensionality , the presence of conservation laws is a relevant feature . for binary fluids ,
to which we restrict our attention , the concentration of the species is conserved as the phase - separation proceeds . the initial stage of the scaling regime , before hydrodynamic effects become relevant , is the subject of this paper . for viscous fluid
this time window can be sufficiently wide to be detected experimentally . in this condition
the phase - separation process is ruled by the evaporation , diffusion and subsequent recondensation of monomers @xcite and the exponent @xmath8 is known to be @xmath10 , independent on the space dimension @xmath11 .
the spatial properties are encoded into the scaling function @xmath12 or , equivalently , into its fourier transform @xmath13 . for large @xmath14
, @xmath15 has a power - law tail of the form @xmath16 : this is the famous porod s law @xcite , long recognized as arising from sharp domains walls .
another relevant parameter affecting the scaling properties is the number of components @xmath17 of the order parameter @xmath18 . for binary fluids
@xmath19 is a scalar field .
most of the analytical techniques developed to study the kinetics , among which the bray - humayun scheme that is considered throughout this article , are however suited for vectorial systems with large-@xmath17 .
let us briefly discuss the main features of vectorial systems that will be recalled in this paper . for systems with a continuous symmetry ,
rotations of the order parameter on the @xmath20 symmetric manifold of the local ground state of the hamiltonian , instead of evaporation and condensation , is the important growth mechanism .
this changes @xmath8 to @xmath21 , regardless both of @xmath11 and @xmath17 ( apart from exceptional cases @xcite ) . concerning the large-@xmath14 behavior of @xmath15 a generalization of the porod s law ,
@xmath22 is obeyed @xcite for @xmath23 .
this form relies on the existence of stable localized topological defects .
much less is known about the tails of @xmath15 when these defects are not present , for @xmath24 .
numerical simulations @xcite found a stretched - exponential decay @xmath25 $ ] . despite this rather simple phenomenology
the problem of deriving the scaling properties from first principles is still open . on the analytical side ,
the only soluble model @xcite for conserved order parameter is the one with @xmath26 . when considering this unphysical limit one hopes that the gross features of the kinetic process are retained for @xmath26 leaving further refinements to @xmath27 perturbation theories .
however , although the global behavior of the @xmath26 model is qualitatively resemblant @xcite of real systems , the scaling symmetry ( [ sca ] ) is violated already at a qualitative level .
the explicit solution of the model @xcite shows in fact that instead of the standard scaling ( [ sca ] ) one has a _ multiscaling _ symmetry produced by the existence of two logarithmically different lengths .
actually the ordering process with @xmath26 is more a _
condensation _ in momentum space @xcite , than a genuine phase - separation process .
these results shadow the possibility to develop a successful theory of the segregation process by systematic @xmath27 expansions around the large-@xmath17 limit . in this scenario a prominent role is played by a class of approximate theories known as gaussian auxiliary field theories , originally introduced by mazenko @xcite .
the essence of these is a non - linear mapping between the order parameter @xmath28 and an auxiliary field @xmath29 .
the actual form of the relationship @xmath30 is largely arbitrary and , depending on different physically motivated choices , different schemes have been developed @xcite .
the basic idea is that with a proper choice of @xmath30 the equation obeyed by the auxiliary field can be treated to lowest order in a mean - field approximation while the basic non - linearities of the evolution are still retained through the non - linear mapping @xmath30 .
ad hoc _ assumption is generally uncontrolled and its use is justified _ a posteriori _ by the satisfactory quality of the predictions .
starting from the method of mazenko , bray and humayun ( bh ) have derived @xcite a non - linear closed equation of motion for the correlation function @xmath31 within the framework of the @xmath27 perturbative expansion .
letting @xmath32 in the bh equation one recovers the large-@xmath17 model previously discussed ; the bh approach , however , does not amount to a straightforward @xmath27-expansion because all orders in @xmath27 are reintroduced through the gaussian auxiliary field approximation . on one side , the uncontrolled character is the limit of the bh scheme , since it is not clear how to improve the approximation , but on the other side , represents a powerful tool to overcome the shortcomings of standard perturbation theory previously discussed . as shown by bh
their equation reinstates standard scaling ( [ sca ] ) as a truly asymptotic symmetry , leaving to multiscaling a merely pre - asymptotic role .
an additional feature at variance with the @xmath26 limit is the form of the scaling function @xmath13 , which , in the bh approximation falls exponentially at large momenta , as opposed to the quartic exponential decay of the @xmath33 model . in this paper
we are interested in the diffusive phase separation process of a binary fluid in the presence of an applied shear flow , namely an imposed fluid velocity along @xmath34 with a constant gradient along @xmath35 .
the presence of the flow radically changes the behavior and even the phenomenology is much less understood .
a first trivial effect is the anisotropic deformation of the growing pattern that appears greatly elongated along @xmath34 @xcite .
then , when dealing with growth laws one has to specify the length of the equilibrated regions for each direction , @xmath36 , and in general @xmath37 .
a more subtle point , that may have important consequences , is the following : stretching of the domains causes ruptures of the network @xcite that may render the segregation incomplete .
actually , the characteristic lengths in the different directions could keep on growing indefinitely , as without shear , or due to domains break - up the system may eventually enter a stationary state characterized by domains with a finite thickness .
in general there is no agreement on this point both from the experimental @xcite and theoretical point of view . on the theoretical side large-@xmath17 calculations @xcite show the existence of a multiscaling regime with ever growing lengths in all directions .
however numerical simulations @xcite can not still establish a clear evidence due to finite size and discretization effects . another intriguing feature which is probably related to the stretching and break - up of the domains is the observation of an oscillatory pattern in the observables .
let us consider the excess viscosity @xmath38 as an example .
stretching of domains require work against surface tension leading to an increase of @xmath38 .
however when a domain breaks up the stress is released since the fragments are more isotropic ; this produces a decrease of @xmath38 .
if repeated in time this mechanism produces an oscillatory pattern that is observed in simulations @xcite and possibly in experiments @xcite . in the large-@xmath17 limit
it can be shown that this process has a periodicity in @xmath39 .
similar patterns are also reported in apparently different contexts as fractures in heterogeneous solids @xcite or in stock market indices @xcite , suggesting a possible common underlying interpretation .
however we do not have nowadays analytical tools to enlighten this feature .
the spatial properties of the mixture are encoded into the correlation function @xmath31 or , equivalently , into its fourier transform @xmath40 , the structure factor .
the latter is directly measured with scattering techniques and is more suited for comparison with the experiments .
the form and the evolution of @xmath41 are themselves modified by the presence of the flow .
when the shear becomes effective , after an initial isotropic regime where @xmath41 has the shape of a circular volcano , the tip of @xmath41 is deformed into an ellipses .
this is expected as a consequence of the anisotropy , and is generally observed in experiments . in some cases
the presence of four pronounced peaks on the edge of @xmath41 can be detected @xcite .
the existence of these multiple maxima has been also observed in numerical studies @xcite of phase separation and in different physical systems such as stationary microemulsion in shear flow @xcite , suggesting that this is a rather general feature of sheared fluids .
a peak in the structure factor is generally interpreted as the signature of a characteristic length proportional to the inverse of the wavevector where the peak is located . in this anisotropic case to each peak one associates one typical length for each space direction . taking into account the symmetry
@xmath42 this suggests that a couple of characteristic lengths for each direction is present : it is not completely clear , however , to which extent this interpretation can be pushed . in this
partly unclear scenario the development of theoretical tools for reliable predictions to be tested in experiments is important . in the case without shear the bh scheme turns out to be a reference theory capable to describe the main ingredients of the phenomenology . therefore in this paper
we address a numerical solution of the bh equation in the presence of shear flow
. we will compare its behavior with previous results pertaining to the large-@xmath17 limit or to numerical simulations of the fully non - linear scalar model .
our results confirm the occurrence of a scaling regime with growing lengths in every direction .
the dynamical exponents are found to coincide with those of the @xmath26 model .
the four peak pattern for the structure factor is exhibited , similarly to what observed both for @xmath33 and in the scalar model , the actual form of @xmath41 being , however , rather different in the three cases .
concerning the oscillating behavior of the typical lengths and of the excess viscosity , on the other hand , the bh model behaves quite differently , in that these are strongly inhibited or even absent , at variance with @xmath26 and @xmath43 .
this paper is divided in 4 sections : in sec .
[ model ] we introduce the bh equation and discuss some numerical details about its integration ; in sec .
[ results ] the results of the numerical solution are presented and a comparison with different approaches is proposed . in sec .
[ conclusions ] we draw the conclusions of this work .
we consider a system described by an order parameter with @xmath17 components .
the evolution is described by the convection - diffusion equation @xcite that generalizes the cahn - hilliard - cook equation in the case of an applied velocity field @xmath44 + = ^2 + * * ( x , t ) [ chc ] where @xmath45 is the vector order parameter that in the scalar case represents the concentration difference between the two species of the fluid , @xmath46 is the space coordinate , @xmath47 is a transport coefficient and @xmath48 is a gaussian stochastic field with expectations @xmath49 describing thermal fluctuations .
the symbol @xmath50 means an ensemble average .
@xmath51 is the equilibrium ginzburg - landau free energy ( \ { * * } ) = dx [ freen ] where @xmath52 .
the parameter @xmath53 distinguishes between the mixed state with @xmath54 ( @xmath55 ) and the phase separated states with @xmath56 .
for a plane shear flow the velocity term is given by v = y e_x [ shear ] where @xmath57 is the unit vector in the flow direction @xmath34 and @xmath58 is the shear rate . in the approximation framework developed by bh @xcite the following equation of motion
can be derived for the equal time correlation function @xmath59 + y = -2 ^ 2\ { ^2 g(r , t)-r(t ) } [ bh ] where we have dropped the component indices due to internal symmetry . in deriving eq .
( [ bh ] ) from eq .
( [ chc ] ) we have let @xmath60 and @xmath61 since this simply amounts to a redefinition of time and space scales and we have neglected the thermal disturbance @xmath62 because it can be shown @xcite that the temperature is an irrelevant parameter below the coexistence line . the quantity @xmath63 is a function of time that is asymptotically determined by the requirement @xmath64 .
here we use the self - consistent determination of @xmath63 r(t)=g[g(0,t)-s_eq ] [ r ] that is usually adopted @xcite . for @xmath26 , namely dropping the cubic term on the r.h.s . , eq . ( [ bh ] ) becomes the @xmath26 model .
we have considered the evolution of eq .
( [ bh ] ) with @xmath43 .
some comments about the role of @xmath17 are in order .
( [ bh ] ) is derived in an @xmath27 perturbative framework .
then , in principle , the quality of the approximation is expected to be better for large @xmath17 and , in any case , this whole approach is meaningful for vectorial systems with continuous symmetry . given that the nature of the symmetry is relevant for the scaling properties , as discussed in sec .
[ intro ] , one can not pretend to push this scheme down to the physically relevant case with @xmath43 in a completely successful way .
however , besides numerical simulation that are quite difficult and not particularly instructive in this case , most of the understanding of phase separation with shear comes nowadays from the @xmath26 model , which surely suffers of much more profound shortcomings than the present approach , if used to infer the properties of physical systems .
the bh scheme represents a first step beyond @xmath26 , despite of course being not conclusive .
it must be recalled , moreover , that at least without shear the bh equation substantially reproduces the behavior of the system with @xmath26 , including multiscaling , in a preasymptotic time domain @xmath65 after which the cubic correction in eq .
( [ bh ] ) becomes effective @xcite .
the choice @xmath43 , besides being appropriate for physical systems , is also the most efficient in order to amplify as much as possible the role of finite-@xmath17 corrections , given the limited timescale numerically accessible .
( [ bh ] ) have been solved on a two dimensional lattice of size @xmath66 , the lattice constant being @xmath67 , via a finite difference first order euler scheme with @xmath68 .
periodic boundary conditions have been adopted .
the initial condition @xmath69 , @xmath70 being a constant , corresponds to a quench from an equilibrium state at infinite temperature .
the parameter of the free energy we use are @xmath71 ; different choices correspond to a redefinition of the order parameter amplitude .
we will present data relative to the case @xmath72 and @xmath73 ; we have tested that other choices produce similar results . the structure factor
@xmath74 obeys the equation -k_x
= -2k^2 c(k , t)-2r(t)d(k , t ) [ ck ] where @xmath75 is the fourier transform of @xmath76 . due to the structure of this equation , however , it is more efficient to compute @xmath5 from eq .
( [ bh ] ) and to obtain @xmath41 by fourier transform . from the knowledge of @xmath41 one
can extract the characteristic lengths along @xmath34 and @xmath35 as @xmath77 and rheological indicators such as the shear stress _
xy ( t)= k_x k_y c(k , t ) [ stress ] for a steady flow the excess viscosity is defined in terms of the shear stress as @xmath78 .
the anisotropic character of the evolution , with the ordered regions aligning along the flow direction , is reflected in the behavior of the correlation function @xmath5 , shown in fig .
( [ figgr ] ) .
@xmath5 has a peak at @xmath79 and is continuously stretched along the @xmath34-direction , assumining increasingly eccentric elliptical patterns .
snapshots of the structure factor at the same times are shown in the @xmath80 plane in fig .
( [ figck ] ) . initially , not shown here
, @xmath41 develops the form of a circular volcano , as in the case without shear .
a later observation , at @xmath81 , shows a deep in the volcano profile along the direction @xmath82 .
this deep becomes progressively more pronounced until @xmath41 appears separated into two symmetric foils . at times of order @xmath83 in each foil a couple of maxima are built up .
we denote by the letter a the peak that in the figure at time @xmath83 is higher . with respect to the other maximum , denoted by b ,
a is characterized by having @xmath84 , whereas in b @xmath85 .
the relative height of these maxima changes in time .
this is observable in fig .
( [ figck ] ) at time @xmath86 : now b is prevailing .
up to this time the qualitative evolution of the structure factor is comparable to what reported for @xmath33 @xcite and for the numerical solution @xcite of the full model equation with @xmath43 .
computer limitations prevent the observation of the system on longer times . only in the @xmath33 case
the dynamics can be followed by numerical integration of the equations on larger timescales and it can be shown that the repeated prevalence of either a or b maxima continues cyclically in time .
it turns also out that the recurrent dominance of a and b is periodic in the logarithm of the strain @xmath87 .
a physical interpretation of this pattern in terms of stretching and break - up of domains is proposed in @xcite .
the spherical average c_sf(k , t)=n^-1dc(k , t ) , [ sf ] where @xmath88 and @xmath89 is the number of lattice points contained in a circular shell of width @xmath90 centered around @xmath91 , and the averages along @xmath92 or @xmath93 c_x(k_y , t)=l^-1dk_x c(k , t ) [ cx ] and c_y(k_x , t)=l^-1dk_y c(k , t ) [ cy ] are plotted in fig .
( [ figmedie ] ) .
this figure shows that in the large-@xmath94 tail the structure factor decays exponentially , as already observed without imposed flow @xcite , at variance with the faster decay observed for @xmath26 @xcite .
as already mentioned , the absence of power - law tails is related to the absence of stable localized topological defects in the bray - humayun approximation .
next we consider the growth laws . for @xmath33 , in the large time region
@xcite @xmath95 and @xmath96 keep increasing with ( logarithmically corrected ) power laws @xmath97 , @xmath98 modulated by a log - periodic oscillation .
the short time part of this pattern is observed in fig .
( [ figradii]b ) .
the same growth exponents are expected for the bh model ( but without logarithmic corrections ) ; this is because the model amounts to an @xmath27 perturbative expansion and , therefore , it is accurate for vectorial systems whose growth exponents belong to the same universality class of @xmath33 .
actually the value of the exponents can be easily determined by means of a scaling analysis .
if standard scaling holds the structure factor can be cast as c(k , t)=l_x(t)l_y(t)f(x , y ) [ scalc ] where @xmath99 and @xmath100 and @xmath15 is a scaling function . from ( [ scalc ] )
one also has d(k , t ) = b l_x(t)l_y(t)h(x , y ) [ scald ] where @xmath101 is a constant and @xmath102 another scaling function . inserting expressions ( [ scalc],[scald ] ) into the equation of motion of @xmath41 ( [ ck ] ) and assuming that asymptotically @xmath95 prevails over @xmath96 one obtains @xmath103y^2 + \frac{l_y(t)^5}{l_x(t)}\gamma x\frac{\partial f(x , y)}{\partial y } \nonumber \\ & - & \frac{1}{4}\frac{dl_y(t)^4}{dt } \left \ { 1+\frac{\frac{d \left [ \ln l_x(t ) \right ] } { dt } } { \frac{d\left [ \ln l_y(t)\right ] } { dt } } \right \ } \label{inferscal}\end{aligned}\ ] ] the l.h.s . of eq .
( [ inferscal ] ) does not depend explicitly either on time and @xmath58 . then in order to fulfill this equation for each @xmath104 both @xmath105 and @xmath58 must drop out from the r.h.s .
then one has \ { ll l_x(t)~t^ + l_y(t)~t^ + r(t)~t^- .
[ exponents ] as already anticipated the exponents are those pertaining to systems with continuous symmetry .
furthermore one can observe that the shear rate enters linearly in @xmath95 but the growth in the shear direction @xmath35 is unaffected by @xmath58 . for completeness
we recall that in the full scalar model one expects different exponents , namely @xmath106 , @xmath107 from scaling @xcite or renormalization group @xcite arguments , but these are not yet clearly observed in simulations .
moreover the very existence of an asymptotic scaling dynamics is debated in this case .
( [ figradii]a ) shows that initially both @xmath95 and @xmath96 start growing with an exponent which is consistent with the value @xmath108 as in absence of shear . in this time regime
the fluid essentially does not feel the presence of the flow , as already discussed for the structure factor .
then , starting from @xmath109 , @xmath96 keeps growing unexpectedly larger than @xmath95 untill at @xmath110 it decreases . up to this point the bh model roughly resembles the @xmath33 case , despite the fact that the overshoot of @xmath96 is much more pronounced in the latter case . fig .
( [ figradii]c ) , instead , shows that the full scalar model behaves quite differently , with a pronounced decrease of @xmath96 in the range of times where in the other models a fast growth was observed . for longer times
the asymptotic stage is entered .
the numerical integration of the bh model presented insofar shows that eq .
( [ exponents ] ) is obeyed but , at variance with the other two cases , the oscillatory pattern is absent or , at least , strongly depressed .
now we turn to the discussion of fig .
( [ figexc ] ) where the behavior of the excess viscosity is reported .
in all the cases considered @xmath111 initially grows to a maximum , then falls negative and raises to a second maximum .
this pattern has been observed also in experiments @xcite and is referred to as _
double overshoot_. after the second maximum the asymptotic stage is entered . on the basis of scaling arguments
one expects that @xmath111 scales as @xmath112 times the inverse of the typical volume @xmath113 of the domains of the two species .
then @xmath114 for vector order parameter and @xmath115 for the scalar model .
the former exponent is observed in fig .
( [ figexc]a ) for the bh model , although in the very last part of the simulation . for @xmath33 the same exponent ( with logarithmic corrections )
has been exactly computed @xcite and observed numerically on larger timescales , but modulated by log - time periodic oscillations @xcite . in the scalar case a definite exponent
can not be extracted from simulations , but the oscillations are observed .
in this article we have considered the behavior of a binary fluid quenched below the mixing temperature in the presence of an applied shear flow .
the description of this system has been carried out in the context of the continuum convection - diffusion equation based on the ginzburg - landau equilibrium free energy .
the nature of our approach , which neglects hydrodynamic effects , limits the domain of applicability of the model to the diffusive regime which takes place at short times .
we have studied this model in the approximation scheme introduced by bray and humayun .
this amounts to an expansion in @xmath27 on top of the auxiliary field theory approximation originally introduced by mazenko .
a comparison is also presented with two different approaches : the @xmath33 limit and the full scalar model , which is the most appropriate for fluids .
the first has the great advantage to be fully soluble analytically , al least in the large time domain , providing an unambiguous description .
furthermore it can be easily studied numerically from the instant of the quench onwards . however , besides the fact that the case @xmath33 is far from physical , it has also proved to violate the qualitative symmetry of standard scaling in favor of multiscaling , rendering this model insecure as an approximation to real systems . on the other hand
the full scalar model is well suited for describing fluids but it can not be studied analytically .
numerical simulations , although important , have not yet provided clear evidences of the basic phenomenology , due to technical limitations .
the bh model is ideally located in between these two : it shares with the large-@xmath17 limit the property of being a closed equation for an observable , the correlation function , that is already an ensemble average . then , even if one resorts in the end to a numerical solution , one still has the great advantage to avoid the average over a large number of realizations .
secondly , due to its perturbative character , it is believed to represent a step forward with respect to the case @xmath33 .
this statement has been proven to be correct at least without shear .
given the essence of the bh scheme , one does not expect to obtain a fully reliable approximation to real systems with @xmath43 .
the spirit of this work is more to establish if and to which extent its phenomenology compares better to the numerical simulations of the physical model than to the case @xmath33 . from the analysis of the behavior of the characteristic lengths and of the excess viscosity
we can conclude that the bh model behaves more like the @xmath33 case with damped oscillations than like the full scalar model , despite that , with the choice @xmath43 , the corrections to the case @xmath33 contained in eq .
( [ bh ] ) are effective almost from the beginning , as it is clear from figs .
( [ figradii],[figexc ] ) .
this is true not only for the value of the dynamic exponents , but also for the qualitative behavior of the typical lengths . on the other hand ,
regarding the oscillating character of most observables , the bh model behaves rather differently from both @xmath33 and @xmath43 in that strong oscillations are observed in the latter two while they are practically absent in the former . under this respect
the bh model is atypical and can not be regarded as a bridge between @xmath33 and @xmath43 .
it is possible that the bh scheme resembles more the behavior of vectorial system with a finite @xmath17 under shear , but this is a totally unexplored field . regarding the important issue of the existence of an asymptotic scaling regime characterized by ever growing lengths @xmath116 , in the time range accessed in this work we do not see any saturation effect and both @xmath95 and @xmath96 keep firmly growing with the expected power laws . in the bh scheme ,
then , the presence of an asymptotic scaling regime is confirmed while a stationary state with domains of a finite thickness is not observed .
bray _ adv .
phys . _ * 43 * , 357 , ( 1994 ) ; j.d .
gunton , m. san miguel and p.s .
sahni , in _ phase transitions and critical phenomena _ , edited by c. domb and j.l .
lebowitz ( academic press , london 1983 ) , vol.8 .
a. coniglio , p. ruggiero and m. zannetti _ phys .
* e 50 * , 1046 ( 1994 ) ; u. marini bettolo marconi and f. corberi , _ europhys .
_ * 30 * , 349 , ( 1995 ) ; s. glotzer and a. coniglio , _ phys . rev . _ * e 50 * , 4241 , ( 1994 ) ; f. corberi and c. castellano , _ phys
. rev . _ * e 58 * , 4658 ( 1998 ) ; c. castellano and f. corberi , _ phys .
rev . _ * e 57 * , 672 ( 1998 ) . c. yeung , y. oono and a.shinozaki , _ phys .
* e 49 * , 2693 , ( 1994 ) ; s. de siena and m. zannetti , _ phys .
* e 50 * , 2621 , ( 1994 ) ; t. otha , d. jasnow and k. kawasaky , _ phys .
lett . _ * 49 * , 1223 , ( 1982 ) ; a.j .
bray and k. humayun , _ phys .
* e 48 * , r1609 ( 1993 ) .
t. hashimoto , k. matsuzaka , e. moses and a. onuki , _ phys .
lett . _ * 74 * , 126 , ( 1995 ) ; t. takebe , f. fujioka , r. sawaoka and t. hashimoto , j. chem . phys . * 98 * , 717 , ( 1993 ) ; j. luger , c. laubner and w. gronsky , _ phys .
lett . _ * 75 * , 3576 , ( 1995 ) ; c.k .
chan , f. perrot and d. beysens , _ phys .
rev . _ * a 43 * , 1826 , ( 1991 ) .
z. laufer , h.l .
jalink and a.j .
staverman , _ journal of polymer science _ * 11 * , 3005 , ( 1973 ) ; s. mani , m.f .
malone , h.h .
winter , j.l .
halary and l. monnerie , _ macromolecules _ * 24 * , 5451 , ( 1991 ) ; s. mani , m.f .
malone and h.h .
winter , _ macromolecules _ * 25 * , 5671 , ( 1992 ) . | the phase separation process which follows a sudden quench inside the coexistence region is considered for a binary fluid subjected to an applied shear flow .
this issue is studied in the framework of the convection - diffusion equation based on a ginzburg - landau free energy functional in the approximation scheme introduced by bray and humayun [ _ phys.rev.lett . _ * 68 * , 1559 , ( 1992 ) ] .
after an early stage where domains form and shear effects become effective the system enters a scaling regime where the typical domains sizes @xmath0 , @xmath1 along the flow and perpendicular to it grow as @xmath2 and @xmath3 .
the structure factor is characterized by the existence of four peaks , similarly to previous theoretical and experimental observations , and by exponential tails at large wavevectors . #
1#2 # # 1##2#2##1 |
in markov chain monte carlo ( mcmc ) , one simulates a markov chain and uses sample averages to estimate corresponding means of the stationary distribution of the chain .
mcmc has become a staple tool in the physical sciences and in bayesian statistics . when sampling the markov chain , the transitions are driven by a stream of independent @xmath0 random numbers .
in this paper , we study what happens when the i.i.d .
@xmath0 random numbers are replaced by deterministic sequences , or by some dependent @xmath0 values . the motivation for replacing i.i.d .
@xmath0 points is that carefully stratified inputs may lead to more accurate sample averages .
one must be cautious though , because as with adaptive mcmc @xcite , the resulting simulated points do not have the markov property .
the utmost in stratification is provided by quasi - monte carlo ( qmc ) points .
there were a couple of attempts at merging qmc into mcmc around 1970 , and then again starting in the late 1990s .
it is only recently that significant improvements have been reported in numerical investigations .
for example , tribble @xcite reports variance reductions of several thousand fold and an apparent improved convergence rate for some gibbs sampling problems .
those results motivate our theoretical work .
they are described more fully in the literature survey below . to describe our contribution ,
represent mcmc sampling via @xmath1 for @xmath2 , where @xmath3 is a nonrandom starting point and @xmath4 .
the points @xmath5 belong to a state space @xmath6 .
the function @xmath7 is chosen so that @xmath5 form an ergodic markov chain with the desired stationary distribution @xmath8 when @xmath9 independently . for a bounded continuous function @xmath10 ,
let @xmath11 and @xmath12 .
then @xmath13 as @xmath14 . in this paper , we supply sufficient conditions on @xmath7 and on the deterministic sequences @xmath15 so that @xmath16 holds when those deterministic sequences are used instead of random ones .
the main condition is that the components of @xmath15 be taken from a completely uniformly distributed ( cud ) sequence , as described below .
ours are the first results to prove that deterministic sampling applied to mcmc problems on continuous state spaces is consistent . in practice
, of course , floating point computations take place on a large discrete state space . but invoking finite precision does not provide a satisfying description of continuous mcmc problems . in a finite state space argument ,
the resulting state spaces are so big that vanishingly few states will ever be visited in a given simulation .
then if one switches from @xmath17 to @xmath18 to @xmath19 bit representations , the problem seemingly requires vastly larger sample sizes , but in reality is not materially more difficult . to avoid using the finite state shortcut ,
we adopt a computational model with infinite precision . as a side benefit ,
this paper shows that the standard practice of replacing genuine i.i.d .
values @xmath15 by deterministic pseudo - random numbers is consistent for some problems with continuous state spaces .
we do not think many people doubted this , but neither has it been established before , to our knowledge .
it is already known from roberts , rosenthal and schwartz @xcite that , under certain conditions , a geometrically ergodic markov chain remains so under small perturbations , such as rounding . that work does not address the replacement of random points by deterministic ones that we make here .
there have been a small number of prior attempts to apply qmc sampling to mcmc problems .
the first appears to have been chentsov @xcite , whose work appeared in 1967 , followed by sobol @xcite in 1974 .
both papers assume that the markov chain has a discrete state space and that the transitions are sampled by inversion .
unfortunately , qmc does not usually bring large performance improvements on such unsmooth problems and inversion is not a very convenient method .
chentsov replaces i.i.d .
samples by one long cud sequence , and this is the method we will explain and then adapt to continuous problems .
sobol uses what is conceptually an @xmath20 matrix of values from the unit interval .
each row is used to make transitions until the chain returns to its starting state .
then the sampling starts using the next row .
it is like deterministic regenerative sampling .
sobol shows that the error converges as @xmath21 in the very special case where the transition probabilities are all rational numbers with denominator a power of @xmath22 .
these methods were not widely cited and , until recently , were almost forgotten , probably due to the difficulty of gaining large improvements in discrete problems , and the computational awkwardness of inversion as a transition mechanism for discrete state spaces .
the next attempt that we found is that of liao @xcite in 1998 .
liao takes a set of qmc points in @xmath23^d$ ] shuffles them in random order , and uses them to drive an mcmc .
he reports 4- to 25-fold efficiency improvements , but gives no theory .
an analysis of liao s method is given in @xcite .
later , chaudary @xcite tried a different strategy using qmc to generate balanced proposals for metropolis
hastings sampling , but found only small improvements and did not publish the work .
craiu and lemieux @xcite also consider multiple - try metropolis and find variance reductions of up to @xmath24% , which is still modest .
earlier , lemieux and sidorsky @xcite report variance reduction factors ranging from about @xmath25 to about @xmath26 in some work using qmc in conjunction with the perfect sampling method of propp and wilson @xcite . only recently have there been significantly large benefits from the combination of qmc and mcmc .
those benefits have mainly arisen for problems on continuous state spaces .
@xcite best results come from gibbs sampling problems computing posterior means . for problems with @xmath27 parameters
, he used every @xmath27-tuple from a small custom built linear feedback shift register ( lfsr ) .
one example is the well - known model used by gelfand and smith @xcite for failure events of @xmath28 pumps from the article by gaver and omurcheartaigh @xcite .
there are @xmath29 unknown parameters , one for each pump and one for the scale parameter in the distribution of pump failure rates .
a second example is a @xmath30 parameter probit model for vasorestriction based on a famous data set from @xcite and analyzed using latent variables as in albert and chib @xcite . of those @xmath30 parameters , the @xmath31 regression coefficients are of greatest interest and @xmath32 latent variables
are nuisance variables .
table [ tab : tribbles ] sets out variance reduction factors found for randomized cud versus i.i.d . sampling .
the improvements appear to grow with @xmath33 , and are evident at very small sample sizes .
@ld3.0d4.0d3.0d4.0d4.0d5.0@ & & & + & & & + * data * & & & & & & + pumps & 286 & 1543 & 304 & 5003 & 1186 & 16089 + vasorestriction & 14 & 15 & 56 & 76 & 108 & 124 + there is another line of research in which large improvements have been obtained by combining qmc with mcmc .
this is the array - rqmc method described in lecuyer , lecot and tuffin @xcite and other articles .
that method simulates numerous chains in parallel using quasi - monte carlo to update all the chains .
it requires a complicated method to match the update variables for each step to the various evolving chains .
this method has achieved variance reductions of many thousand fold on some problems from queuing and finance .
very few properties have been established for it , beyond the case of heat particles in one dimension that was considered by morokoff and caflisch @xcite . finally , jim propp s rotor - router method is a form of deterministic markov chain sampling .
it has brought large efficiency improvements for some problems on a discrete state space and has been shown to converge at better than the monte carlo rate on some problems .
see , for example , doerr and friedrich @xcite .
the use of cud sequences that we study has one practical advantage compared to the rotor - router , array - rqmc , regenerative sampling , and the other methods .
it only requires replacing the i.i.d . sequence used in a typical mcmc run by some other list of numbers .
the paper is organized around our main results which appear in section [ sec : consistency ] .
theorem [ thm : withhomecouple ] gives sufficient conditions for consistency of qmc - mcmc sampling by metropolis
theorem [ thm : ifmalsfuh ] gives sufficient conditions for consistency of qmc - mcmc sampling for the systematic scan gibbs sampler .
section [ sec : background ] contains necessary background and notation for the two main theorems of section [ sec : consistency ] .
it introduces quasi - monte carlo and markov chain monte carlo giving key definitions we need in each case .
that section presents the rosenblatt chentsov transformation .
we have combined a classic sequential inversion method based on the rosenblatt transformation with an elegant coupling argument that chentsov @xcite used .
the consistency results for metropolis hastings ( theorem [ thm : withhomecouple ] ) make moderately strong assumptions in order to ensure that a coupling occurs .
section [ sec : homeandcouple ] shows that those assumptions are satisfied by some metropolized independence samplers and also by some slice samplers .
we also assumed some riemann integrability properties for our mcmc proposals .
the riemann integral is awkward compared to the lebesgue integral , but considering it is necessary when we want to study specific algorithms on deterministic inputs .
section [ sec : riemaninteg ] gives sufficient conditions for an mcmc algorithm to satisfy the required riemann integrability conditions .
our consistency results for the gibbs sampler ( theorem [ thm : ifmalsfuh ] ) require some contraction properties and some jordan measurability .
section [ sec : gibbsexample ] shows that these properties hold under reasonable conditions .
section [ sec : openclose ] has a brief discussion on open versus closed intervals for uniform random numbers .
our conclusions are in section [ sec : conclusions ] .
the lengthier or more technical proofs are placed in the .
our random vectors are denoted by @xmath34 for @xmath35 .
points in the unit cube @xmath23^d$ ] are denoted by @xmath36 .
two points @xmath37 with @xmath38 for @xmath39 define a rectangle @xmath40 $ ] , denoted by @xmath41 $ ] for short .
the indicator ( or characteristic ) function of a set @xmath42 is written @xmath43 .
we assume the reader is familiar with the definition of the ( proper ) riemann integral , for a bounded function on a finite rectangle @xmath41\subset{\mathbb{r}}^d$ ] .
the bounded set @xmath42 is jordan measurable if @xmath43 is riemann integrable on a bounded rectangle containing @xmath44 . by lebesgue s theorem ( see section [ sec : riemaninteg ] ) @xmath44 is jordan measurable if @xmath45 . here
@xmath46 denotes lebesgue measure on @xmath47 , and @xmath48 is the boundary of @xmath44 , that is , the set on which @xmath43 is discontinuous . here , we give a short summary of quasi - monte carlo .
further information may be found in the monograph by niedereiter @xcite .
qmc is ordinarily used to approximate integrals over the unit cube @xmath23^d$ ] , for @xmath49 .
let @xmath50^d$ ] . the qmc estimate of @xmath51^d } f({\mathbf{x}})\,{{d}}{\mathbf{x}}$ ] is @xmath52 , just as we would use in plain monte carlo .
the difference is that in qmc , distinct points @xmath5 are chosen deterministically to make the discrete probability distribution with an atom of size @xmath53 at each @xmath5 close to the continuous @xmath54^d$ ] distribution . the distance between these distributions is quantified by discrepancy measures .
the local discrepancy of @xmath55 at @xmath56^d$ ] is @xmath57 the star discrepancy of @xmath55 in dimension @xmath27 is @xmath58^d}}|\delta({\mathbf{a}};{\mathbf{x}}_1,\ldots,{\mathbf{x}}_n)|.\ ] ] for @xmath59 , the star discrepancy reduces to the kolmogorov smirnov distance between a discrete and a continuous uniform distribution .
a uniformly distributed sequence is one for which @xmath60 as @xmath14 .
if @xmath5 are uniformly distributed then @xmath61 provided that @xmath62 is riemann integrable . under stronger conditions than riemann integrability
, we can get rates of convergence for qmc .
the koksma hlawka inequality is @xmath63 is the total variation of @xmath62 in the sense of hardy and krause . for properties of @xmath64 and other multidimensional variation measures ,
see @xcite . equation ( [ eq : khbound ] ) gives a deterministic upper bound on the integration error , and it factors into a measure of the points quality and a measure of the integrand s roughness .
there exist constructions @xmath55 where @xmath65 holds for any @xmath66 .
therefore , functions of finite variation can be integrated at a much better rate by qmc than by mc .
rates of convergence of @xmath67 , where @xmath68 denotes the smoothness of the integrand which can therefore be arbitrarily large , can also be achieved @xcite .
equation ( [ eq : khbound ] ) is not usable for error estimation .
computing the star discrepancy is very difficult @xcite , and computing @xmath69 is harder than integrating @xmath62 .
practical error estimates for qmc may be obtained using randomized quasi - monte carlo ( rqmc ) . in rqmc each @xmath70^d$ ] individually while the ensemble @xmath55 has @xmath71 for some @xmath72 . for an example , see @xcite .
a small number of independent replicates of the rqmc estimate can be used to get an error estimate .
rqmc has the further benefit of making qmc unbiased . for a survey of rqmc ,
see @xcite .
a key distinction between qmc and mc is that the former is effective for riemann integrable functions , while the latter , in principle , works for lebesgue integrable functions . in practice
, mc is usually implemented with deterministic pseudo - random numbers .
the best generators are proved to simulate independent @xmath54 $ ] random variables based on either discrepancy measures over rectangles or on spectral measures .
those conditions are enough to prove convergence for averages of riemann integrable functions , but not for lebesgue integrable functions . as a result ,
ordinary monte carlo with pseudo - random numbers is also problematic for lebesgue integrable functions that are not riemann integrable . in the markov chain context
, we need a lesser known qmc concept as follows .
a sequence @xmath73 $ ] is completely uniformly distributed ( cud ) if for any @xmath74 the points @xmath75 satisfy @xmath76 as @xmath14 .
this is one of the definitions of a random sequence from knuth @xcite , and it is an important property for modern random number generators . using a cud sequence in an mcmc is akin to using up the entire period of a random number generator , as remarked by niederreiter @xcite in 1986 .
it is then necessary to use a small random number generator .
the cud sequences used by tribble @xcite are miniature versions of linear congruential generators and feedback shift register generators . as such ,
they are no slower than ordinary pseudo - random numbers . in the qmc context
, we need to consider nonoverlapping @xmath27-tuples @xmath77 for @xmath78 .
it is known @xcite that @xmath79 in the qmc context , the function @xmath62 subsumes all the necessary transformations to turn a finite list of i.i.d .
@xmath54 $ ] random variables into the desired nonuniformly distributed quantities , as well as the function of those quantities whose expectation we seek . in some problems
, we are unable to find such transformations , and so we turn to mcmc methods .
suppose that we want to sample @xmath80 for a density function @xmath8 defined with respect to lebesgue measure on @xmath81 . for definiteness , we will seek to approximate @xmath82 . in this section ,
we briefly present mcmc . for a full description of mcmc , see the monographs by liu @xcite or robert and casella @xcite . in an mcmc simulation ,
we choose an arbitrary @xmath83 with @xmath84 and then for @xmath78 update via @xmath85 where @xmath86^d$ ] and @xmath7 is an update function described below .
the distribution of @xmath5 depends on @xmath87 only through @xmath88 and so these random variables have the markov property . the function @xmath7 is chosen so that the stationary distribution of @xmath5 is @xmath8
. then we estimate @xmath89 by @xmath90 as before . if a burn - in period was used , we assume that @xmath3 is the last point of it .
first , we describe the metropolis
hastings algorithm for computing @xmath91 from the current point @xmath92 and @xmath93^d$ ] .
it begins with a proposal @xmath94 taken from a transition kernel @xmath95 . with genuinely random proposals ,
the transition kernel gives a complete description .
but for either quasi - monte carlo or pseudo - random sampling , it matters how we actually generate the proposal .
we will assume that @xmath96 @xmath54 $ ] random variables are used to generate @xmath94 via @xmath97
. then the proposal @xmath94 is either accepted or rejected with probability @xmath98 .
the decision is typically based on whether the @xmath27th random variable @xmath99 is below @xmath44 .
the function @xmath100^d\to{\mathbb{r}}^s$ ] is a generator for the distribution @xmath101 on @xmath102 if @xmath103 when @xmath104^d$ ] . for @xmath92 ,
let @xmath105^{d-1}\to{\omega}$ ] be a generator for the transition kernel @xmath95 with conditional density . the metropolis
hastings sampler has @xmath106 where @xmath107 and @xmath108 the mis update is a special case of the metropolis
hastings update in which @xmath109 does not depend on @xmath110 .
the rwm update is a special case of the metropolis
hastings update in which @xmath111 for some generator @xmath112 not depending on @xmath110 .
let @xmath113 with @xmath114 and @xmath115 . to construct the systematic scan gibbs sampler ,
let @xmath116 be a @xmath117-dimensional generator of the full conditional distribution of @xmath118 given @xmath119 for all @xmath120 .
this gibbs sampler generates the new point using @xmath93^d$ ] .
write @xmath121 with @xmath122^{k_j}$ ] .
the systematic scan gibbs sampler has @xmath123 where , for @xmath124 , @xmath125}}({\mathbf{u}}_j)\ ] ] and @xmath126 } = ( \phi_1({\mathbf{x}},{\mathbf{u}}),\ldots,\phi_{j-1}({\mathbf{x}},{\mathbf{u } } ) , x_{j+1},\ldots , x_d)$ ] .
let @xmath8 be a probability density function on @xmath127 .
let @xmath128 and let @xmath129 be the uniform distribution on @xmath130 .
the inversive slice sampler is the systematic scan gibbs sampler for @xmath129 with each @xmath131 using inversion for every @xmath132}}$ ] .
there are many other slice samplers .
see @xcite .
it is elementary that @xmath133 implies @xmath80 .
it is more usual to use @xmath134 , but our setting simplifies when we assume @xmath135 is updated first .
we generate our random variables as functions of independent uniform random variables .
the generators we consider require a finite number of inputs , so acceptance - rejection is not directly covered , but see the note in section [ sec : conclusions ] .
for an encyclopedic presentation of methods to generate nonuniform random vectors , see devroye @xcite . here , we limit ourselves to inversion and some generalizations culminating in the rosenblatt chentsov transformation introduced below .
we will not need to assume that @xmath8 can be sampled by inversion .
we only need inversion for an oracle used later in a coupling argument .
let @xmath101 be the cdf of @xmath136 , and for @xmath137 define @xmath138 take @xmath139 and @xmath140 , using extended reals if necessary .
then @xmath141 has distribution @xmath101 on @xmath142 when @xmath143 $ ] .
multidimensional inversion is based on inverting the rosenblatt transformation @xcite .
let @xmath101 be the joint distribution of @xmath144 .
let @xmath145 be the marginal cdf of @xmath146 and for @xmath147 , let @xmath148 be the conditional cdf of @xmath118 given @xmath149 .
the inverse rosenblatt transformation @xmath150 of @xmath93^s$ ] is @xmath151 where @xmath152 and @xmath153 if @xmath104^s$ ] , then @xmath154 .
we will use the inverse rosenblatt transformation as a first step in a coupling argument which extends the one in chentsov @xcite .
let @xmath150 be the inverse rosenblatt transformation for the stationary distribution @xmath8 and let @xmath7 be the update function for mcmc .
the rosenblatt chentsov transformation of the finite sequence @xmath155^d$ ] is the finite sequence @xmath156 , with @xmath157 , where @xmath158 and @xmath159 for @xmath160 .
the rosenblatt chentsov transformation starts off using @xmath161 and inversion to generate @xmath3 and then it applies whatever generators are embedded in @xmath7 with the innovations @xmath15 , to sample the transition kernel .
the transition function @xmath7 need not be based on inversion .
in this section , we prove sufficient conditions for some deterministic mcqmc samplers to sample consistently .
the same proof applies to deterministic pseudo - random sampling .
first , we define consistency , then some regularity conditions , and then we give the main results .
our definition of consistency is that the empirical distribution of the mcmc samples converges weakly to @xmath8 .
[ def : consist ] the triangular array @xmath162 for @xmath33 in an infinite set @xmath163 consistently samples the probability density function @xmath8 if @xmath164 holds for all bounded continuous functions @xmath165 .
the infinite sequence @xmath166 consistently samples @xmath8 if the triangular array of initial subsequences with @xmath167 for @xmath2 does . in practice
, we use a finite list of vectors and so the triangular array formulation is a closer description of what we do .
however , to simplify the presentation and avoid giving two versions of everything , we will work only with the infinite sequence version of consistency .
triangular array versions of cud sampling for discrete state spaces are given in @xcite .
it suffices to use functions @xmath62 in a convergence - determining class .
for example , we may suppose that @xmath62 is uniformly continuous @xcite , or that @xmath168}$ ] @xcite .
when @xmath8 is a continuous distribution , we may use @xmath169}$ ] . here
, we define some assumptions that we need to make on the mcmc update functions .
[ def : coupleregion ] let @xmath170^d$ ] have positive jordan measure . if @xmath171 implies that @xmath172 for all @xmath173 , then @xmath174 is a coupling region .
consider two iterations @xmath175 and @xmath176 with the same innovations @xmath15 but possibly different starting points @xmath3 and @xmath177 .
if @xmath178 , then @xmath179 holds for all @xmath180 . in section [ sec : homeandcouple ] , we give some nontrivial examples of mcmc updates with coupling regions .
[ def : regular ] let @xmath181 be the last point generated in the rosenblatt chentsov transformation , viewed as a function on @xmath23^{d(m+1)}$ ] . the mcmc is _ regular _ ( _ for bounded continuous functions _ )
if the function @xmath182 is riemann integrable on @xmath23^{d(m+1)}$ ] whenever @xmath62 is bounded and continuous . note that if an mcmc is regular , then the definition of the rosenblatt chentsov transformation implies that @xmath183^{d(m+1 ) } } f({\mathbf{x}}_m({\mathbf{u}}_0,\ldots,{\mathbf{u}}_m ) ) \,{{d}}{\mathbf{u}}_0 \cdots{{d}}{\mathbf{u}}_m = \int_{{\omega } } f({\mathbf{x } } ) \pi({\mathbf{x } } ) \,{{d}}{\mathbf{x}}\ ] ] for any @xmath184 and all bounded continuous functions @xmath62 .
we can , of course , define regularity for mcmc also with respect to other classes of functions . indeed ,
there are numerous equivalent conditions for regularity .
for example , the portmanteau theorem ( @xcite , chapter 1.2 ) implies that it is enough to assume that the functions @xmath62 are bounded and uniformly continuous .
of interest are also indicator functions of rectangles since they appear in the definition of the local discrepancy at ( [ eq : localdis ] ) .
the following theorem states some equivalent conditions . to simplify the statements
, we write that mcmc is _
regular for indicator functions _ whenever @xmath185 is riemann integrable on @xmath23^{d(m+1)}$ ] , where @xmath44 is either @xmath186 $ ] with @xmath187 finite or @xmath188 .
[ thm : eq_regularity ] the following statements are equivalent : a. mcmc is regular for bounded continuous functions .
b. mcmc is regular for bounded uniformly continuous functions . c. mcmc is regular for indicator functions @xmath189}$ ] of rectangles
@xmath41 $ ] .
this result follows by applying the portmanteau theorem ( @xcite , chapter 1.2 ) and some methods from real analysis .
a regular mcmc is one that satisfies any ( and hence all ) of the above .
theorem [ thm : withhomecouple ] below is the main result that we will use for metropolis
hastings sampling .
one does not expect cud sampling to correct for an mcmc algorithm that would not be ergodic when sampled with i.i.d .
ergodicity is assured through our assumption that there is a coupling region .
section [ sec : homeandcouple ] below shows that some nontrivial mcmc methods have such regions .
theorem [ thm : withhomecouple ] does not require the detailed balance condition that metropolis
hastings satisfies , and so it may apply to some nonreversible chains too .
[ thm : withhomecouple ] let @xmath190 and let @xmath191 , and for @xmath192 let @xmath193 where @xmath7 is the update function of a regular mcmc with a coupling region @xmath174 . if @xmath194 for a cud sequence @xmath195 , then @xmath55 consistently samples @xmath8 .
the proof of theorem [ thm : withhomecouple ] is in the .
it shows that the fraction of points @xmath5 in a bounded rectangle @xmath41 $ ] converges to @xmath196}\pi({\mathbf{x}})\,{{d}}{\mathbf{x}}$ ] .
almost the same proof technique applies to expectations of bounded continuous functions .
the gibbs sampler can be viewed as a special case of metropolis
hastings with acceptance probability one .
however , it is more straightforward to study it by applying results on iterated function mappings to ( [ eq : defphi ] ) using methods from diaconis and freedman @xcite and alsmeyer and fuh @xcite . in this subsection
, we assume that @xmath197 is a complete separable metric space .
we assume that the update function @xmath91 is jointly measurable in @xmath110 and @xmath198 and that it is lipschitz continuous in @xmath110 for any @xmath198 .
lipschitz continuity is defined through the metric @xmath199 on @xmath200 .
the lipschitz constant , which depends on @xmath198 , is @xmath201 for each @xmath202^d$ ] , define @xmath203 .
next , we present a theorem from alsmeyer and fuh @xcite on iterated random mappings .
the @xmath33 step iteration , denoted @xmath204 , is defined by @xmath205 and for @xmath206 .
[ thm : ifmalsfuh ] let the update function @xmath91 be jointly measurable in @xmath110 and @xmath198 with @xmath207^d } \log(\ell({\mathbf{u}}))\,{{d}}{\mathbf{u}}<0 $ ] and , for some @xmath208 , @xmath207^d } \ell({\mathbf{u}})^p\,{{d}}{\mathbf{u}}<\infty$ ] .
assume that there is a point @xmath209 with @xmath207^d } \log^+ ( d(\phi({\mathbf{x}}',{\mathbf{u}}),{\mathbf{x}}'))\,{{d}}{\mathbf{u}}<\infty$ ] and @xmath210 .
then there is a @xmath211 such that for all @xmath212 there is a @xmath213 such that for every @xmath214 @xmath215 this follows by specializing corollary 2.5(a ) of @xcite to the present setting .
[ thm : withiterfunction ] let @xmath216 be a complete separable metric space and let @xmath217 be a cud sequence such that for every sequence @xmath218 of natural numbers with @xmath219 , we have @xmath220 .
let @xmath221 , and for @xmath192 let @xmath193 be the gibbs sampler update for stationary distribution @xmath8 .
assume that @xmath7 satisfies the conditions of theorem [ thm : ifmalsfuh ] and that there is a @xmath212 such that @xmath222^{dm}\dvtx d(\phi_m({\mathbf{x}},{\mathbf{v } } ) , \phi_m(\widehat{{\mathbf{x}}},{\mathbf{v } } ) ) > \gamma^m \}\ ] ] is jordan measurable for all @xmath223 and @xmath224 .
under these conditions , if the gibbs sampler is regular , then @xmath225 consistently samples @xmath8 .
the proof of theorem [ thm : withiterfunction ] is in the .
like theorem [ thm : withhomecouple ] , it shows that bounded rectangles @xmath41 $ ] have asymptotically the correct proportion of points .
once again , similar arguments apply for bounded continuous functions of @xmath110 .
although not explicitly stated there , the proof of @xcite , theorem 1 , shows the existence of sequences @xmath217 for which @xmath226 for all @xmath227 , where @xmath228 is a constant independent of @xmath33 and @xmath27 . unfortunately
, no explicit construction of such a sequence is given in @xcite .
then for any sequence @xmath229 of natural numbers with @xmath219 we obtain that @xmath230 in theorem [ thm : withhomecouple ] , we assumed that the coupling region @xmath174 is jordan measurable in theorem [ thm : withiterfunction ] , we do not have a coupling region , but still have an analogous assumption , namely that the sets @xmath231 are jordan measurable . a condition on @xmath7 which guarantees that @xmath231 is jordan measurable is given in section [ sec : gibbsexample ] .
theorem [ thm : withhomecouple ] used coupling regions .
these are somewhat special .
but they do exist for some realistic mcmc algorithms .
let @xmath7 be the update for the metropolized independence sampler on @xmath81 obtaining the proposal @xmath232 , where @xmath112 generates samples from the density @xmath233 , which are accepted when @xmath234 assume that the importance ratio is bounded above , that is , @xmath235 suppose also that there is a rectangle @xmath41\subset[0,1]^{d-1}$ ] of positive volume with @xmath236 } \frac{\pi(\psi({\mathbf{u}}))}{p(\psi ( { \mathbf{u } } ) ) } > 0.\ ] ] then @xmath237\times[0,\eta/\kappa]$ ] is a coupling region .
the set @xmath174 has positive jordan measure .
suppose that @xmath238 .
then @xmath239 and so @xmath240 , regardless of @xmath110 .
[ lem : slicecouple ] let @xmath8 be a density on a bounded rectangular region @xmath241\subset{\mathbb{r}}^s$ ] .
assume that @xmath242 holds for all @xmath92 .
let @xmath243\times[0,\kappa]$ ] be the domain of the inversive slice sampler .
let @xmath244 for @xmath86^{s+1}$ ] be the update for the inversive slice sampler and put @xmath245 . if @xmath246\times[0,1]^s$ ] , then @xmath247 . if @xmath248 , then @xmath249 and @xmath250 are in the set @xmath251 $ ] .
the distribution of @xmath110 given @xmath135 for any @xmath252 $ ] is @xmath253 $ ] .
therefore , @xmath254 ( componentwise ) .
lemma [ lem : slicecouple ] does not couple the chains because @xmath255 and @xmath256 are different in general .
but because @xmath247 , a coupling will happen at the next step , that is , @xmath257 when @xmath258\times[0,1]^s$ ] . one could revise theorem [ thm : withhomecouple ] to include couplings that happen within some number @xmath259 of steps after @xmath238 happens . in this case
, it is simpler to say that the chain whose update comprises two iterations of the inversive slice sampler satisfies theorem [ thm : withhomecouple ] . for a chain
whose update is just one iteration , the averages over odd and even numbered iterations both converge properly and so that chain is also consistent .
alternatively , we could modify the space of @xmath135 values so that all @xmath252 $ ] are identified as one point .
then @xmath174 is a coupling region .
the result of lemma [ lem : slicecouple ] also applies to slice samplers that sample @xmath260 and then @xmath261 using an @xmath262-dimensional generator that is not necessarily inversion .
theorem [ thm : withhomecouple ] proves that mcmc consistently samples @xmath8 when implemented using cud sequences .
we required certain riemann integrability conditions in defining regular rosenblatt
chentsov transformations . here
, we verify that nontrivial mcmc algorithms can have regular rosenblatt chentsov transformations .
it seems odd to use the riemann integral over @xmath263 years after lebesgue @xcite .
but pseudo - random number generators are now typically designed to meet an equidistribution criterion over rectangular regions @xcite .
other times they are designed with a spectral condition in mind .
this again is closely related to riemann integrability via the weyl @xcite condition where @xmath16 for all trigonometric polynomials @xmath264 if and only if @xmath55 are uniformly distributed . unless one is using physical random numbers , the riemann integral , or perhaps the improper riemann integral is almost implicit .
a function from @xmath42 to @xmath102 for is riemann integrable if all of its @xmath262 components are . to study how riemann integrability propagates
, we will use the following two definitions . for a function @xmath265 ,
the discontinuity set of @xmath62 is @xmath266 if @xmath62 is only defined on @xmath267 , then @xmath268 where @xmath269 for @xmath270 and @xmath271 for @xmath272 . for a function @xmath265 ,
the graph of @xmath62 is @xmath273 lebesgue s theorem , next , provides a checkable characterization of riemann integrability .
[ thm : lebesgues ] let @xmath42 be bounded and let @xmath274 be a bounded function .
then @xmath62 is riemann integrable iff @xmath275 .
see marsden and hoffman @xcite , page 455 . here
, we show that riemann integrability adds a special requirement to the way an algorithm is implemented .
then we give an example to show that propagation rules for riemann integrability are more complicated than are those for continuity and differentiability .
suppose that @xmath101 is the @xmath276 distribution for some @xmath277 .
if we take @xmath278 and @xmath279 then we find that @xmath280 is discontinuous only on a set of measure zero .
it is trivially bounded , and these two facts imply it is riemann integrable on @xmath23 ^ 2 $ ] .
another transformation for the same distribution @xmath101 is @xmath281 and @xmath282 changing the conditional distribution of @xmath283 given @xmath146 on a set of measure @xmath284 leaves the distribution @xmath101 of @xmath110 unchanged . but for this version , we find @xmath62 can be discontinuous on more than a set of measure @xmath284 and so this inverse rosenblatt transformation of @xmath101 is not regular . in practice , of course , one would use the regular version of the transformation . but
propagating riemann integrability to a function built up from several other functions is not always straightforward .
the core of the problem is that the composition of two riemann integrable functions need not be riemann integrable . as an example @xcite ,
consider thomae s function on @xmath285 , @xmath286 where it is assumed that @xmath233 and @xmath287 in the representation @xmath288 have no common factors .
this @xmath62 is continuous except on @xmath289 and so it is riemann integrable .
the function @xmath290 is also riemann integrable . but
@xmath291 for @xmath292 , which is famously not riemann integrable .
the class of riemann integrable functions , while more restrictive than we might like for conclusions , is also too broad to use in propagation rules . first , we show that the acceptance - rejection step in metropolis
hastings does not cause problems with riemann integrability .
[ lem : accrej ] let @xmath293 and suppose that @xmath294 , @xmath295 and @xmath44 are real - valued riemann integrable functions on @xmath23^k$ ] . for @xmath93^{k+1}$ ] define @xmath296 then @xmath62 is riemann integrable on @xmath23^{k+1}$ ] .
first , @xmath297)\cup g(a)$ ] .
riemann integrability of @xmath294 gives @xmath298 .
similarly , @xmath299 .
therefore , @xmath300)=0 $ ] . turning to @xmath301
, we split the domain @xmath23^k$ ] of @xmath44 into @xmath302 congruent subcubes @xmath303 ( whose boundaries overlap ) . then @xmath304 $ ] , where @xmath305 and @xmath306 . as a result @xmath307 .
riemann integrability of @xmath44 implies this upper bound vanishes as @xmath14 .
therefore , @xmath308 and so @xmath309 and the result follows by lebesgue s theorem . in the mcmc context ,
@xmath294 and @xmath295 are the @xmath310th component of the proposal and the previous state , respectively , @xmath44 is the acceptance probability , and @xmath198 is the ensemble of uniform random variables used in @xmath311 stage rosenblatt chentsov coupling and @xmath312 . for consistency results , we study the proportion of times @xmath313 $ ] .
it is enough to consider the components one at a time and in turn to show @xmath314 and @xmath315 are riemann integrable .
however , as the example with thomae s function shows , even the indicator function of an interval applied to a riemann integrable function can give a non - riemann integrable composite function .
we may avoid truncation by employing bounded continuous test functions .
we will use the following simple corollary of lebesgue s theorem .
[ lem : ctsofriemann ] for @xmath316 and @xmath317 , let @xmath318 be riemann integrable functions from @xmath23^k$ ] to a bounded interval @xmath319\subset{\mathbb{r}}$ ] .
let @xmath295 be a continuous function from @xmath319^k$ ] to @xmath142 .
then @xmath320 is riemann integrable on @xmath23^k$ ] .
because @xmath295 is continuous , @xmath321 . but
therefore , @xmath322 and so @xmath62 is riemann integrable by lebesgue s
. we can also propagate riemann integrability through monotonicity . if @xmath294 is a monotone function from @xmath142 to @xmath142 and @xmath62 is the indicator of an interval , then @xmath323 is the indicator of an interval too , and hence is riemann integrable , when that interval is of finite length .
[ lem : rosenblatt ] let @xmath324 be the cdf of @xmath146 and for @xmath147 , let @xmath325 be the conditional cdf of @xmath118 given @xmath326 .
suppose that the cdfs @xmath325 are continuous functions of @xmath327 and that the quantile functions @xmath328 are continuous in @xmath329\times{\mathbb{r}}^{j-1}$ ] , for @xmath147 .
define functions @xmath330 and @xmath331 for @xmath147 , where @xmath332 .
then for , the set @xmath333 is jordan measurable . by hypothesis , @xmath334 is a continuous function of @xmath93^s$ ] , for @xmath335 , and so is @xmath336 .
this latter only depends on @xmath337 , for @xmath338 , and so we write it as @xmath339 . for @xmath335 ,
let @xmath340 .
the set @xmath341 is the interval @xmath342 $ ] , and hence is jordan measurable .
suppose @xmath343 is jordan measurable for @xmath344 .
then @xmath345 ) \cap g_{k+1 } \qquad \mbox{where } g_{k+1 } = \bigl\ { { \mathbf{u}}_{1\dvtx ( k+1)}\mid u_{k+1 } \le g_{k+1 } ( { \mathbf{u}}_{1\dvtx k})\bigr\}.\end{aligned}\ ] ] the set @xmath346 $ ] is jordan measurable because @xmath343 is .
the boundary of @xmath347 is contained within the intersection of the graph of @xmath348 and the boundary of @xmath23^{k+1}$ ] and so @xmath347 is jordan measurable .
the result follows by induction because @xmath349 . here , we give sufficient conditions for the rosenblatt chentsov transformation to be regular . for integer @xmath350 ,
let @xmath351 be the endpoint of the rosenblatt chentsov transformation of @xmath23^{(d+1)m}$ ] , started with a riemann integrable function @xmath150 and continued via the metropolis
hastings update @xmath7 .
let @xmath7 be defined in terms of the proposal function @xmath352^{d-1}\to{\mathbb{r}}^s$ ] with proposal density @xmath353 and target density @xmath354 .
let @xmath62 be a bounded continuous function on @xmath102 . if @xmath112 is bounded and @xmath94 , @xmath355 and @xmath8 are bounded continuous functions , then @xmath182 is a riemann integrable function of the variables @xmath23^{(d+1)m}$ ] used in the rosenblatt chentsov transformation .
we only need to show that @xmath356 is a riemann integrable function of @xmath357^{d(m+1)}$ ] and then the result follows by lemma [ lem : ctsofriemann ] .
we proceed by induction . for @xmath358
, @xmath359 is bounded and continuous on @xmath23^d$ ] , hence it is riemann integrable . now suppose that @xmath360 is a riemann integrable function on @xmath23^{dm}$ ] .
let @xmath361 be the value @xmath360 , written as a riemann integrable function on @xmath23^{dm+d-1}$ ] , so it ignores its last @xmath96 arguments .
let @xmath362 be the proposal @xmath363 , @xmath364 .
this is a continuous function @xmath365 of two riemann integrable functions on @xmath23^{d(m+1)-1}$ ] and so it is riemann integrable .
next , @xmath366 is a continuous function of both @xmath360 and @xmath367 which are in turn riemann integrable functions on @xmath23^{dm+d-1}$ ] , and so @xmath366 is riemann integrable .
then @xmath351 is a riemann integrable function on @xmath23^{dm+d}$ ] , by lemma [ lem : accrej ] , completing the induction .
in studying the gibbs sampler , we made several assumptions .
first , we required jordan measurability for the sets @xmath368 .
second , we required a contraction property . in this section , we show that those assumptions are reasonable .
we give an example where the conditions of theorem [ thm : withiterfunction ] are satisfied , that is , the sets @xmath369 are jordan measurable for all @xmath370 and @xmath214 ( for some suitable domain @xmath6 ) .
assume ( additionally to the assumptions made in theorem [ thm : withiterfunction ] ) that @xmath91 is totally differentiable with continuous derivative with respect to @xmath198 for each @xmath371 and that @xmath27 is based on the @xmath372 norm for some @xmath373 .
further , assume that the gradient of @xmath374 with respect to @xmath198 vanishes only on a null set for all @xmath375 , @xmath376 , that is , @xmath377^{d}\dvtx
\nabla_{{\mathbf{u}}}\ , d(\phi({\mathbf{x}},{\mathbf{u}}),\phi(\widehat{{\mathbf{x}}},{\mathbf{u } } ) ) = { \mathbf{0}}\}\bigr ) = 0,\ ] ] for all @xmath378 , @xmath379 , where @xmath380 denotes the lebesgue measure and where @xmath381 denotes the gradient .
then , for all @xmath223 , we also have @xmath377^{dm}\dvtx \nabla_{{\mathbf{u}}}\ , d(\phi_m({\mathbf{x}},{\mathbf{u}}),\phi_m(\widehat{{\mathbf{x}}},{\mathbf{u } } ) ) = { \mathbf{0}}\}\bigr ) = 0\ ] ] for all @xmath214 , @xmath376 .
let @xmath382 with @xmath376 be fixed .
then for almost all @xmath383^{dm}$ ] we have @xmath384 .
therefore , there is a @xmath385 such that @xmath386 for all @xmath387 , where @xmath388^{dm}\dvtx \|{\mathbf{u}}^\ast-{\mathbf{v}}\|_{l_2 } < \delta\}$ ] is a neighborhood of @xmath389 .
therefore , the directional derivative at a point @xmath387 is different from @xmath284 , except on a hyperplane , that is , almost everywhere .
hence , by the mean value theorem , the function @xmath390 for @xmath391 can at most be constant on a hyperplane , which has lebesgue measure @xmath284 .
note that @xmath392 , therefore there is a countable number of elements @xmath393 and numbers @xmath394 with the properties of @xmath389 and @xmath395 described above and for which we have @xmath396^{dm}$ ] .
therefore , we have @xmath377^{dm}\dvtx d(\phi_m({\mathbf{x}},{\mathbf{u}}),\phi_m(\widehat{{\mathbf{x}}},{\mathbf{u } } ) ) = c \ } \bigr ) = 0,\ ] ] for any @xmath397 .
the set of points where @xmath398 is discontinuous is given by @xmath399^{dm}\dvtx \forall\delta > 0\ \exists{\mathbf{v}},{\mathbf{v } } ' \in n_\delta({\mathbf{u } } ) \mbox { such that } \\ & & \hspace*{4.5pt } d(\phi_m({\mathbf{x}},{\mathbf{v}}),\phi_m(\widehat{{\mathbf{x}}},{\mathbf{v } } ) ) > \gamma^m \mbox { and } d(\phi_m({\mathbf{x}},{\mathbf{v}}'),\phi_m(\widehat{{\mathbf{x}}},{\mathbf{v } } ' ) ) \le \gamma^m \}.\end{aligned}\ ] ] as @xmath369 and @xmath400^{dm}\dvtx d(\phi_m({\mathbf{x}},{\mathbf{u}}),\phi_m(\widehat{{\mathbf{x}}},{\mathbf{u } } ) ) < \gamma^m \}$ ] are open , it follows that @xmath401^{dm}\dvtx d(\phi_m({\mathbf{x}},{\mathbf{u}}),\phi_m(\widehat{{\mathbf{x}}},{\mathbf{u } } ) ) = \gamma^m \}.\ ] ] therefore , @xmath402 and lebesgue s theorem ( see theorem [ thm : lebesgues ] ) implies that @xmath369 is jordan measurable . here , we illustrate how the gibbs sampler yields a contraction for the probit model . in this model , @xmath403 and @xmath404 for @xmath2 for independent @xmath405 .
the coefficient @xmath406 has a noninformative prior distribution .
the predictors are @xmath407 .
we define the matrix @xmath408 with @xmath409 element @xmath410 .
we assume that @xmath408 has rank @xmath233 .
the state of the markov chain is @xmath411 , where @xmath412 .
given the observed data @xmath413 , we can use the gibbs sampler to simulate the posterior distribution of @xmath414 and @xmath415 .
a single step of the gibbs sampler makes the transition @xmath416 for @xmath316 using generators given explicitly below .
the values @xmath417 are the components of @xmath418 .
we also write the transitions as @xmath419 where @xmath7 and its components @xmath420 and @xmath421 ( for @xmath414 and @xmath422 , resp . )
are given explicitly below .
given @xmath414 , the components of @xmath422 are independent , with @xmath423 we may generate them from @xmath424 by @xmath425 given @xmath422 , the distribution of @xmath414 is @xmath426 . we may generate it using @xmath427 via @xmath428 thus equation ( [ eq : betafromz ] ) defines @xmath420 while ( [ eq : zfrombeta ] ) defines @xmath421 .
the framework in @xcite allows one to pick a metric that conforms to the problem .
we use the metric @xmath429 where @xmath430 and @xmath431 we show below that @xmath432 for pairs @xmath433 of distinct points in @xmath434
. both metrics @xmath435 and @xmath436 are also norms , which simplifies our task .
suppose first that @xmath437 .
then it follows easily that @xmath438 and @xmath439 , so then the left - hand side of ( [ eq : itcontracts ] ) is @xmath284 . as a result , we may assume without loss of generality that @xmath440 . with this assumption , we will use the bound @xmath441\\[-8pt ] & & \qquad\le\max\biggl ( \frac{d_1(\beta^{(k)}-\beta^{\prime(k ) } ) } { d_1(\beta^{(k-1)}-\beta^{\prime(k-1 ) } ) } , \frac{d_2({\mathbf{z}}^{(k)}-{\mathbf{z}}^{\prime(k ) } ) } { d_1(\beta^{(k-1)}-\beta^{\prime(k-1 ) } ) } \biggr).\nonumber\end{aligned}\ ] ] we begin by studying the update to @xmath422 . subtracting @xmath442 from both sides of ( [ eq : zfrombeta ] ) , applying @xmath443 , differentiating with respect to @xmath414 and gathering up terms , we find that @xmath444 where @xmath445 and @xmath446 is the @xmath447 probability density function .
it is clear that @xmath448 .
next , we show that @xmath449 .
we begin by inverting ( [ eq : zfrombeta ] ) to get @xmath450 substituting ( [ eq : ufrombeta ] ) into ( [ eq : lambdafrom ] ) and simplifying yields @xmath451 now consider the function @xmath452 .
this function is nonnegative and decreasing , using a mill s ratio bound from @xcite . when @xmath453 , then @xmath454 because then @xmath455 .
we also used symmetry of @xmath456 . if instead @xmath457 , then @xmath458 because then @xmath459 . either way , @xmath460 and therefore @xmath461 for all @xmath462 .
writing the previous results in a compact matrix form , we have @xmath463 where @xmath464 . similarly , equation ( [ eq : betafromz ] ) yields @xmath465 thus , for the @xmath422 update with any @xmath418 , @xmath466\\[-8pt ] & \le&{\mathop{\sup_{\beta,{\mathbf{z}}}}_{(x\xi)^{\mathsf{t}}x\xi= 1 } } \vert\lambda(\beta,{\mathbf{z } } ) x \xi\vert < 1.\nonumber\end{aligned}\ ] ] for the @xmath414 update , applying the chain rule gives @xmath467 and then @xmath468\\[-8pt ] & = & { \sup_{\beta,{\mathbf{z}},\vert\eta\vert= 1 } } \vert x(x^{\mathsf{t}}x)^{-1}x^{\mathsf{t}}\lambda\eta\vert \nonumber\\ & \le&{\max_{1\le i\le n}\lambda_i } \nonumber\\ & < & 1,\nonumber\end{aligned}\ ] ] using the nonexpansive property of the projection matrix @xmath469 . by combining ( [ eq : zup ] ) with ( [ eq : bup ] ) , we establish the contraction ( [ eq : itcontracts ] ) .
in the lebesgue formulation , @xmath470 and @xmath54^d$ ] are the same distribution , in that they can not be distinguished with positive probability from any countable sample of independent values .
riemann integrals are usually defined for @xmath23^d$ ] and discrepancy measures are usually defined for either @xmath23^d$ ] or @xmath471 .
these latter theories are designed for bounded functions . in monte carlo simulations ,
sometimes values @xmath472 are produced .
these end points can be problematic with inversion , where they may yield extended real values , and hence good practice is to select random number generators supported in the open interval @xmath285 .
for our gibbs sampler example with the probit model , we required @xmath418 .
this was necessary because otherwise the values @xmath91 might fail to belong to @xmath200 .
our slice sampler example had @xmath200 equal to the bounded rectangle @xmath41 $ ] .
then values @xmath472 do not generate sample points outside @xmath200 .
our metropolized independence sampler did not require bounded support .
it could produce extended real values .
those however are not problematic for weak convergence , which is based on averages of @xmath189}({\mathbf{x}}_i)$ ] or other bounded test functions .
also , the chain will not get stuck at an unbounded point .
we have demonstrated that mcqmc algorithms formed by metropolis hastings updates driven by completely uniformly distributed points can consistently sample a continuous stationary distribution .
some regularity conditions are required , but we have also shown that those conditions hold for many , though by no means all , mcmc updates . the result is a kind of ergodic theorem for qmc like the ones in @xcite and @xcite for finite state spaces . when rqmc is used in place of qmc to drive an mcmc simulation , then instead of cud points , we need to use weakly cud points .
these satisfy @xmath473 for all @xmath66 and all @xmath49 .
our version of mcmc above leaves out some methods in which one or more components of @xmath15 are generated by acceptance - rejection sampling because then we can not assume @xmath474 . a modification based on splicing i.i.d .
@xmath54 $ ] random variables into a cud sequence was proposed by liao @xcite and then shown to result in a weakly cud sequence in @xcite .
we do not expect that a global substitution of qmc points will always bring a large improvement to mcmc algorithms .
what we do expect is that means of smooth functions of the state vector in gibbs samplers will often benefit greatly from more even sampling .
it is also a fair question to ask when one needs an mcmc result computed to the high precision that qmc sometimes makes possible .
gelman and shirley @xcite address this issue , distinguishing task 1 ( inference about a parameter @xmath475 ) from task 2 [ precise determination of @xmath476 or more generally @xmath477 conditional on the data , or a posterior quantile of @xmath475 ] .
the accuracy of task 1 problems may be limited more by sample size than by monte carlo effort .
task 2 problems include computation of normalizing constants and problems where one wants to report numerically stable , and hence more reproducible , simulation output .
this appendix contains the lengthier proofs .
we need one technical lemma about cud points .
consider overlapping blocks of @xmath479-tuples from @xmath480 , with starting indices @xmath27 units apart .
if @xmath480 are cud then these overlapping blocks are uniformly distributed .
the proof works by embedding the @xmath479-tuples into nonoverlapping @xmath481-tuples . for large @xmath482 ,
the boundary effect between adjacent blocks becomes negligible .
this result is also needed for the argument in @xcite .
[ lem : tech ] for @xmath483 , let @xmath484 $ ] . for integers @xmath485 , let @xmath486 .
if @xmath487 are completely uniformly distributed , then @xmath488^{dk}$ ] are uniformly distributed .
choose any @xmath489^{dk}$ ] .
let @xmath490 be the volume of @xmath491 .
for integers @xmath317 , define @xmath492 on @xmath23^{rdk}$ ] by @xmath493 .
each @xmath492 has riemann integral @xmath494 .
we use @xmath492 on nonoverlapping blocks of length @xmath481 from @xmath487 : @xmath495 after using ( [ eq : overlapornot ] ) . taking @xmath482
as large as we like , we get @xmath496 .
it follows that @xmath497 for any rectangular subset @xmath498^{dk}$ ]
. therefore , @xmath499 too , for otherwise some rectangle @xmath500 would get too few points .
now , we prove the main theorems from section [ sec : consistency ] .
proof of theorem [ thm : withhomecouple ] pick @xmath501 . now let @xmath502 and for @xmath2 define the sequence @xmath503 as the rosenblatt chentsov transformation of @xmath504 .
suppose that @xmath7 is regular and for a bounded rectangle @xmath41\subset{\mathbb{r}}^s$ ] , let @xmath505}({\mathbf{x}})$ ] .
then @xmath506 where @xmath507 and @xmath508 for @xmath509 , notice that @xmath510 $ ] if and only if @xmath511 lies in a @xmath512-dimensional region @xmath513 .
the region @xmath513 has volume @xmath196 } \pi({\mathbf{x } } ) \,{{d}}{\mathbf{x}}$ ] because @xmath514)$ ] is @xmath196 } \pi({\mathbf{x } } ) \,{{d}}{\mathbf{x}}$ ] when @xmath515^{d(m+1)}$ ] .
it has a riemann integrable indicator function by hypothesis .
then because @xmath217 are cud , and using lemma [ lem : tech ] with @xmath516 , we get @xmath517 now , consider @xmath518 .
the only nonzero terms arise when @xmath519 .
this in turn requires that the coupling region @xmath174 is avoided @xmath311 consecutive times , by @xmath520
. then @xmath521 belongs to a region of volume at most @xmath522 .
choose @xmath311 large enough that @xmath523 . then @xmath524 for the third term , @xmath525 is at most @xmath526 , which goes to @xmath284 as @xmath527 .
thus , we have @xmath528 } - \int_{[{\mathbf{a}},{\mathbf{b } } ] } \pi({\mathbf{x } } ) \,{{d}}{\mathbf{x}}\biggr| < \varepsilon.\ ] ] as @xmath529 was chosen arbitrarily , the result follows for this case .
the result holds trivially for the function @xmath530 , hence we are done .
proof of theorem [ thm : ifmalsfuh ] we use the notation from the proof of theorem [ thm : withhomecouple ] . as in the proof of theorem [ thm : withhomecouple ]
, we write @xmath531 as the sum of three terms .
the first and third terms vanish by the same arguments we used in theorem [ thm : withhomecouple ] . for the second term , we have @xmath532 let @xmath529 be arbitrary .
we show that @xmath533 .
as @xmath529 is arbitrary , this then implies that @xmath534 .
assume that the gibbs sampler is regular for rectangles and for a bounded positive volume rectangle @xmath41 \subset{\mathbb{r}}^s$ ] let @xmath505}({\mathbf{x}})$ ] .
for @xmath535 , let @xmath536 and put @xmath537}$ ] and @xmath538}$ ] . because @xmath539 , the triple @xmath540 must be in the set @xmath541 . likewise @xmath542 . by inspecting all @xmath543 cases in @xmath544 , we find that @xmath545 , for @xmath546 and @xmath547 where @xmath548
. choose @xmath385 such that @xmath549 \setminus[{\mathbf{a}}+{\bolds{\delta}},{\mathbf{b}}-{\bolds{\delta } } ] ) } \pi({\mathbf{x } } ) \,{{d}}{\mathbf{x } } < \frac{\varepsilon}{3}.\ ] ] as the gibbs sampler is regular for rectangles , @xmath217 is a cud sequence , and @xmath550 is constructed using the rosenblatt chentsov transformation we have @xmath551^{dm+d}\dvtx { \mathbf{x}}'_{i , m , m } \in [ { \mathbf{a}}-{\bolds{\delta}},{\mathbf{b}}+{\bolds{\delta}}]\setminus[{\mathbf{a}},{\mathbf{b } } ] \ } ) \\ & & \qquad = \int_{{\omega}\cap ( [ { \mathbf{a}}-{\bolds{\delta}},{\mathbf{b}}+{\bolds{\delta } } ] \setminus[{\mathbf{a}}+{\bolds{\delta}},{\mathbf{b}}+{\bolds{\delta } } ] ) } \pi({\mathbf{x } } ) \,{{d}}{\mathbf{x } } < \frac{\varepsilon}{3},\end{aligned}\ ] ] and so @xmath552 .
the points @xmath550 and @xmath553 have different starting points @xmath554 and @xmath555 , but are updated @xmath311 times using the same @xmath556 , that is , @xmath557 , @xmath558 and @xmath559 .
therefore , theorem [ thm : ifmalsfuh ] implies that there is a constant @xmath228 such that for all sufficiently large @xmath560 the region @xmath561^{dm}\dvtx d(\phi_m({\mathbf{x}}'_{i , m,0},({\mathbf{v}}_1,\ldots,{\mathbf{v}}_m)),\\ & & \hspace*{142.2pt}\phi_m({\mathbf{x}}_{i},({\mathbf{v}}_1,\ldots,{\mathbf{v}}_m ) ) ) > \gamma^m\},\end{aligned}\ ] ] has volume at most @xmath562 .
let @xmath563 .
let @xmath564 if @xmath41 \cap{\omega}= \varnothing$ ] or @xmath565 = \varnothing$ ] and @xmath566 \cap{\omega } , { \mathbf{y } } ' \in{\omega}\setminus[{\mathbf{a}}-{\bolds{\delta}},{\mathbf{b}}+{\bolds{\delta}}]\}$ ] otherwise .
let @xmath567 be such that @xmath568 and @xmath569 .
now take @xmath570 .
for large enough @xmath33 , we can take @xmath571 .
then @xmath572 has volume at most @xmath573 .
thus , @xmath574 implies that @xmath575 , which in turn implies that @xmath576 , and so @xmath577 .
therefore , we have @xmath578 a similar argument shows that @xmath579 .
combining the three bounds yields @xmath580 establishing consistency when the gibbs sampler is regular .
since the result holds trivially for the function @xmath530 , the result follows .
the coupling region in theorem [ thm : withhomecouple ] was replaced by a mean contraction assumption @xmath207^d } \log(\ell({\mathbf{u}}))\,{{d}}{\mathbf{u } } < 0 $ ] in theorem [ thm : withiterfunction ] . this way we obtain ( possibly different )
coupling type regions @xmath581 for each @xmath582 .
we remedy this situation by letting @xmath311 depend on @xmath33 , which in turn requires us to use a stronger assumption on the cud sequence @xmath217 , namely , that @xmath583 .
we thank seth tribble , erich novak , ilya m. sobol and two anonymous reviewers for helpful comments .
gelman , a. and shirley , k. ( 2010 ) .
inference from simulations and monitoring convergence . in _ handbook of markov chain monte carlo : methods and applications_. ( s. brooks , a. gelman , g. jones and x .-
meng , eds . ) 131143 . chapman and hall / crc press , boca raton , fl .
lecuyer , p. and lemieux , c. ( 1999 ) .
quasi - monte carlo via linear shift - register sequences . in _ proceedings of the 1999 winter simulation conference _
( p. a. farrington , h. b. nembhard , d. t. sturrock and g. w. evans , eds . ) 632639 .
ieee press , piscataway , nj . owen , a. b. ( 1995 ) .
randomly permuted @xmath585-nets and @xmath586-sequences . in _
monte carlo and quasi - monte carlo methods in scientific computing _ ( h. niederreiter and p. jau - shyong shiue , eds . ) 299317 .
springer , new york .
owen , a. b. ( 2005 ) .
multidimensional variation for quasi - monte carlo . in
_ contemporary multivariate analysis and design of experiments : in celebration of prof .
kai - tai fang s 65th birthday _ ( j. fan and g. li , eds . ) .
world sci .
hackensack , nj . | the random numbers driving markov chain monte carlo ( mcmc ) simulation are usually modeled as independent @xmath0 random variables .
tribble [ markov chain monte carlo algorithms using completely uniformly distributed driving sequences ( 2007 ) stanford univ . ]
reports substantial improvements when those random numbers are replaced by carefully balanced inputs from completely uniformly distributed sequences . the previous theoretical justification for using anything other than i.i.d .
@xmath0 points shows consistency for estimated means , but only applies for discrete stationary distributions .
we extend those results to some mcmc algorithms for continuous stationary distributions .
the main motivation is the search for quasi - monte carlo versions of mcmc . as a side benefit ,
the results also establish consistency for the usual method of using pseudo - random numbers in place of random ones . , and . |
the h i proximity effect has been widely studied since it was discovered .
one thing of particular interest is that the study of the h i proximity effect provides a method of measuring the intensity of the muvb at the h i edge .
however , the h i proximity effect does not reveal information on the spectral shape of the muvb .
we demonstrate that the proximity effect on metal elements may provide a method for constraining the spectral shape of the muvb .
songalia and cowie ( 1996 ) @xcite reported that metal elements are detected in 75% of clouds with @xmath0 .
the lack of data for studying the proximity effect on metal elements is going to overcome .
if future observations detect what we expect for the proximity effect on metal elements , it will confirm that the proximity effect is an ionization effect and photoionization is the dominant ionization mechanism for metal ions in which the proximity effect is seen .
the electronic configurations of metal elements are complicated and a large number of ionization stages can coexist simultaneously . in general , a simple feature does not exist for metal proximity effect , unlike for the h i proximity effect .
therefore , we may study the dependence of the column density ratios of two ions in successive ionization stages on @xmath1 .
we found that [ o iv / o iii]@xmath2 is the most optimum indicator of the spectral shape of the muvb ( we use [ x]@xmath2 to represent the ratio of the value of x in the vicinity of the qso to that in background case ) .
the hst does provide the possibility to observe o iv and o iii simultaneously @xcite and @xcite .
it is clear from figure 1 that [ o iv / o iii]@xmath2 strongly depends on the assumed spectral shape of the muvb and less sensitive to the assumed @xmath3 ( the ionization parameter in background case ) in small @xmath1 region .
therefore , [ o iv / o iii]@xmath2 is good for constraining the spectral shape even if @xmath3 is unknown .
the ratio si iv / c iv , contrast to previous thought , is not a good indicator to the spectral shape , because it strongly depends on both the ionization parameter @xmath3 and the spectral shape of the muvb . a small uncertainty on either of them must results in a big error in determination of the other .
we also found that the extra ionizing radiation from qsos does not always reduce the ratio c iv / h i , but enhances c iv / h i in the case of @xmath4 .
so , the observed excess of c iv systems in luminous qsos may be a photoionization effect , but not a gravitational lensing effect @xcite .
99 liu , j.m . , & meiksin , a. , 1997 , in preparation songaila , a. , & cowie , l.l .
1996 , aj , 112 , 335 tripp , t.m .
, lu , l. , & savage , b.d .
1997 , [ astro - ph/9703080 ] vanden berk , d.e . ,
quashnock , j.m . , york , d.g . , & yanny , b. 1996 , 469 , 78 vogel , s. , & reimers , d. 1995 , a&a , 294 , 377 | using our photoionization code cirrus @xcite , we show that the study of proximity effect on metal elements provides a possibility of placing constraints on the spectral shape of the metagalactic uv background ( muvb ) . from theoretical point of view , the most optimum indicator of the spectral shape of the muvb is the ratio o iv / o iii . unlike previous thought ,
the ratio si iv / c iv is not a good indicator of the spectral shape because of its strong dependence on an ionization parameter as well .
we also find that the observed excess of c iv systems in luminous qsos may be a photoionization effect , but not a gravitational lensing effect @xcite . |
micro - swimmers are of general interest lately , motivated by both engineering and biological problems @xcite
. they can be remarkably subtle as was illustrated by e. m. purcell in his famous talk on life at low reynolds numbers " @xcite where he introduced a deceptively simple swimmer shown in fig .
[ fig : purcell ] .
purcell asked `` what will determine the direction this swimmer will swim ? ''
this simple looking question took 15 years to answer : koehler , becker and stone @xcite found that the direction of swimming depends , among other things , on the stroke s _ amplitude _ : increasing the amplitudes of certain small strokes that propagate the swimmer to the right result in propagation to the left .
this shows that even simple qualitative aspects of low reynolds number swimming can be quite un - intuitive .
purcell s swimmer made of three slender rods can be readily analyzed numerically by solving three coupled , non - linear , first order , differential equations @xcite . however , at present there appears to be no general method that can be used to gain direct qualitative insight into the properties of the solutions of these equations .
our first aim here is to to describe a geometric approach which allows one to describe the qualitative features of the solution of the swimming differential equations without actually solving them .
our tools are geometric .
the first tool is the notion of curvature borrowed from non - abelian gauge theory @xcite .
this curvature can be represented graphically by landscape diagrams such as figs .
[ fig : optstrkhh],[fig : fphieta=2],[fig : purcellcurvx ] which capture the qualitative properties of general swimming strokes .
we have taken care not to assume any pre - existing knowledge about gauge theory on part of the reader .
rather , we have attempted to use swimming as a natural setting where one can build and develop a picture of the notions of non - abelian gauge fields .
purcell s original question , `` what will determine the direction this swimmer will swim ? ''
can often be answered by simply looking at such landscape pictures .
our second tool is a notion of metric and curvature associated with the dissipation .
the `` dissipation curvature '' can be described as a landscape diagram and it gives information on the geometry of `` shape space '' .
this gives us useful geometric tools that give qualitative information on the solutions of rather complicated differential equations .
we begin by illustrating these geometric methods for the symmetric version of purcell s swimmer , shown in fig .
[ fig : symmetric - p ] .
symmetry protects the swimmer against rotations so it can only swim on a straight line .
this makes it simple to analyze by elementary analytical means . in particular , it is possible to predict , using the landscape portraits of the swimming curvature fig .
[ fig : optstrkhh ] , which way it will swim .
in this ( abelian ) case the swimming curvature gives full quantitative information on the swimmer .
we then turn to the non - abelian case of the usual purcell s swimmer which can also rotate .
there are now several notions of swimming curvatures : the rotation curvature and the two translation curvature .
the translation curvatures are non - abelian .
this means that they give precise information of small strokes but this information can not be integrated to learn about large strokes
. this can be viewed as a failure of stokes integration theorem .
nevertheless , as we shall explain , they do give lots of qualitative information about large strokes as well . .
the location of the swimmer in the plane is determined by two position coordinates @xmath0 and one orientation @xmath1 .
the lengths of each arm is @xmath2 and the length of the body @xmath3.,title="fig:",width=302 ] +
purcell s swimmer , which was invented as the simplest animal that can swim that way " @xcite , is not simple to analyze . a variant of it that is simple to analyze is shown in the fig [ fig : symmetric - p ] . and the length of the body @xmath3 .
the position of the swimmer is denoted by @xmath4 . , title="fig:",width=8 ] + the swimmer has four arms , each of length @xmath5 and one body arm of length @xmath3 , ( possibly of zero length ) .
the swimmer can control the angles @xmath6 and the arms are not allowed to touch .
both angles increase in the counterclockwise direction , @xmath7 .
being symmetric , this swimmer can not rotate and can swim only in the `` body '' direction .
it falls into the class of `` simple swimmers '' which includes the `` three linked spheres '' of najafi and glolestanian @xcite and the pushmepullyou @xcite , whose hydrodynamics is elementary because they can not turn .
let us first address purcell s question `` what will determine the direction this swimmer will swim ? '' for the stroke shown in fig . [
fig : optstrkhh ] . in the stroke ,
the swimmer moves the two arms backwards together and then bring them forward one by one .
the first half of the cycle pushes the swimmer forward and the second half pulls it back .
which half of the cycle wins ? to answer that , one needs to remember that swimming at low reynolds numbers relies more on effective anchors than on good propellers .
since one needs twice the force to drag a rod transversally than to drag it along its axis @xcite , an open arm @xmath8 acts like an anchor .
this has the consequence that rowing with both arms , in the same direction and in phase , is _ less effective _ than bringing them back out of phase .
the stroke actually swims backwards .
this reasoning also shows that the swimmer is sisyphian : it performs a lot of forward and backward motion for little net gain .
the swimming equation at low reynolds number is the requirement that the total force ( and torque ) on the swimmer is zero .
the total force ( and torque ) is the sum of the forces ( and torques ) on the four arms and body . for the symmetric swimmer ,
the torque and force in the transversal direction vanish by symmetry .
the swimming equation is the condition that the force in the body - direction vanishes .
this force depends linearly on the known rate of change of the controls , @xmath9 and the unknown the velocity @xmath10 of the `` body - rod '' .
it gives a linear equation for the velocity .
for slender arms and body , the forces are given by cox @xcite slender body theory : the element of force , @xmath11 , acting on a segment of length @xmath12 located at the point @xmath13 on the slender body is given by @xmath14 where @xmath15 is a unit tangent vector to the slender - body at @xmath13 and @xmath16 its velocity there .
@xmath17 is the viscosity and @xmath18 the slenderness ( the ratio of length to diameter ) .
the force on the a - th arm depends linearly on the velocities of the controls @xmath19 and swimming velocity @xmath10 .
for example , the force component in the x - direction on the a - th arm takes the form _
f_aj^x_j+f_a^xxx where @xmath20 are functions of the controls , given by elementary integrals [ eq : f_jk ] f_a^xx= k ( ^2_a-2 ) _ 0^ds , f_aj^x=2 k_j _ 0^s ds similar equations hold for the left arm and the body .
the requirement that the total force on the swimmer vanishes gives a linear relation between the variation of the controls and the displacement @xmath21 @xmath22 where [ eq : a - def ] a(,)= , = as one expects , the body is just a `` dead weight '' and a trim swimmer with @xmath23 is best .
.,title="fig:",width=453 ] + the notation @xmath21 stresses that the differential displacement does not integrate to a function of the controls , @xmath24 .
this is the essence of swimming : @xmath25 fails to return to its original value with @xmath26 .
for this reason , swimming is best captured not by the differential one - form @xmath21 but by the differential two - form @xmath27 @xmath28 and @xmath29 is the rational function given in eq .
( [ eq : a - def ] ) .
@xmath30 is commonly known as the curvature @xcite and its surface integral for a region enclosed by a curve @xmath31 gives , by stokes , the distance covered in one stroke .
it gives complete information on both the direction of swimming and distance .
the total curvature associated with the full square of shape space , @xmath32 is 0 by symmetry . the total positive curvature associated with the triangular half of the square
, @xmath33 is 0.274 .
this means that swimmer can swim , at most , about a quarter of its arms length in a single stroke .
@xmath30 is a differential two form and as such is assigned a numerical value only when in comes with a region of integration . by itself
, it has no numerical value . to say that the curvature is large at a given point in shape , or control , space requires fixing some a - priori measure .
for example one can pick the flat measure for @xmath26 in which case @xmath34 gives numerical values for the curvature .
however , one can pick instead the flat measure for @xmath35 one gets a different function .
a natural measure on shape space is determined by the dissipation .
we turn to it now .
the power of swimming at low reynolds numbers is quadratic in the driving : @xmath36 where @xmath37 is a function on shape space and we use the summation convention where repeated indices are summed over .
this suggests the natural metric in shape space is , in either coordinate systems , @xmath38 in particular , the associated area form is d_1d_2= d_1d_2 the curvature can now be assigned a natural numerical value [ eq : numerical - curvature ] 2 a^2(,)= 2 a^2(,)_1_2 each arm of the symmetric purcell swimmer dissipate energy at the rate [ eq : dissipation - arm ] -_0^df(s)vds & = & -k_0^ ( ( tv)^2 -2 vv)ds + & = & -k_0^ ( x^2 - 2 s^2 _
j^2 ^2_j-2(s _ j_j - x)^2 ) ds + & = & 3 ( 3x^2 + 2 _ j^2 + 6 _ j x ) and the total energy dissipation by the arms is evidently [ eq : dissipation ] 3 ( 6x^2 + 2 ( _ 1 ^ 2+_2 ^ 2 ) + 6 ( _ 1+_2 ) x ) in a body - less swimmer , @xmath23 , this is also the total dissipation , and we consider this case from now on . since we are interested in the metric up to units , we shall henceforth set . ]
@xmath39 . plugging the swimming equation , eq . (
[ eq : dx ] ) gives @xmath40 : [ eq : metric ] g()=a(,0 ) ( cc 5 - 2_2 ^ 2 + _ 1 ^ 2 & _ 1 _ 2 +
_ 1 _ 2 & 5 - 2_1 ^ 2 + _ 2 ^ 2 + ) , _ j=_j and @xmath41 is given in eq .
( [ eq : a - def ] ) . in particular
, @xmath42 is a smooth function on shape space while @xmath43 is singular at the boundaries .
one can now meaningfully plot the curvature which is shown in fig .
[ fig : optstrkhh ] efficient swimming covers the largest distance for given energy resource and at a given speed .
alternatively , it minimizes the energy needed for covering a given distance at a given speed . fixing the speed for a given distance is equivalent to fixing the time @xmath44 . in this formulation
the variational problem takes the form of a problem in lagrangian mechanics of minimizing the action @xmath45 where q is a lagrange multiplier .
@xmath21 is given in eq .
( [ eq : dx ] ) . this can be interpreted as a motion of a a charged particle on a curved surface in an external magnetic field @xcite .
conservation of energy then says that the solution has constant speed ( in the metric @xmath40 ) . for a closed path ,
the kinetic term is then proportional to the length of the path and the constraint is the flux enclosed by it .
thus the variational problem can be rephrase geometrically as the isoperimetric problem " : find the shortest path that encloses the most flux .
the charged particle moves on a curved surface .
how does this surface look like ? from the dissipation metric we can calculate , using brioschi formula @xcite , the gaussian curvature @xmath46 ( not to be confused with @xmath30 ) of the surface . a plot of it is given in fig .
[ fig : gausscurv ] . + inspection of fig .
[ fig : optstrkhh ] suggest that pretty good strokes are those that enclose only one sign of the curvature @xmath30 .
the actual optimal stroke can only be found numerically .
it is plotted in figure [ fig : optstrkhh ] .
the efficiency for this stroke is about the same as the efficiency of the purcell s swimmer for rectangular strokes of @xcite , but less than the optimally efficient strokes found in @xcite .
cox theory does not allow the arms to get too close .
how close they are allowed to get depends on the slenderness @xmath18 .
the smallest angle allowed @xmath47 must be such that @xmath48 . as the optimal stroke gets quite close to the boundary , with @xmath49 it can be taken seriously only for sufficiently slender bodies with @xmath50 , which is huge .
the optimal stroke is therefore more of mathematical than physical interest .
one can use a refine slender body approximation by taking high order terms in cox s expansion for the force .
this will leave the structure without changes , but will made eq .
[ eq : f - yosi ] and eq .
[ eq : metric ] much more complicated . from a mathematical point of view
it is actually quite remarkable that a minimizer exists . by this
we mean that the optimal stroke does not hit the boundary of shape space @xmath51 where cox theory is squeezed out of existence .
this can be seen from the following argument .
inspection of eq .
( [ eq : numerical - curvature ] ) and eq .
( [ eq : metric ] ) shows that the curvature vanishes linearly near the boundary of shape space ( this is most easily seen in the @xmath35 coordinates ) .
suppose now that the optimal path ran along the boundary .
shifting the path a distance @xmath52 away from the boundary would shorten it linearly in @xmath52 while the change in the flux integral will be only quadratic .
this shows that the path that hits the boundary can not be a minimizer .
purcell swimmer can move in either direction in the plane and can also rotate . since the euclidean group is not abelian ( rotations and translations do not commute ) the notion of `` swimming curvature '' that proved to be so useful in the abelian case needs to be modified .
as we shall explain , landscape figures can be used to give qualitative geometric understanding of the swimming and in particular can be used to answer purcell question `` what will determine the direction this swimmer will swim ? '' . however , unlike the abelian case , the swimming curvature does not give full quantitative information on the swimming and one can not avoid solving a system of differential equations in this case if one is interested in quantitative details . the location and orientation of the swimmer ( in the lab frame ) shall be denoted by the triplet @xmath53 where @xmath1 is the orientation of the swimmer , see fig .
[ fig : purcell ] , and @xmath0 are cartesian coordinates of the center of the `` body . '' .
] we use super - indices and greek letters to designate the response while lower indices and roman characters designate the controls @xmath54 .
the common approach to low reynolds numbers swimming is to write the equations of motion in a fixed , lab frame .
we first review this and then describe an alternate approach where the equations of motions are written in a frame that instantaneously coincides with the swimmer . by general principles of low reynolds numbers hydrodynamics
there is a linear relations between the change in the controls increases counterclockwise and @xmath55 clockwise . ]
@xmath56 and the response @xmath57 [ eq : linear - response ] -.1em.8ex-.6emdx^=a^_j d_j , ( summation over repeated indices implied . ) note that @xmath58 since there are two controls , while @xmath59 for the three responses .
the response coefficients @xmath60 are functions of both the control coordinates @xmath61 and the location coordinates @xmath62 of the swimmer in the lab .
however , in a homogeneous medium it is clear that @xmath63 can only be a function of the orientation @xmath1 .
moreover , in an isotropic medium it can only dependent on the orientation @xmath64 through [ eq : ra ] ^_i ( , ) = r^ ( ) a^_i ( ) ; r^ ( ) = ( * 20c 1 & 0 & 0 + 0 & & - + 0 & & + ) in the lab frame , the nature of the solution of the differential equations is obscured by the fact that one can not determine @xmath57 independently for different points on the stroke ( because of the dependence on @xmath65 ) .
the coefficients @xmath66 may be viewed as the transport coefficients in a rest frame that instantaneously coincides with the swimmer .
they play a key role in the geometric picture that we shall now describe . in the frame of the swimmer one
has [ eq : linear - response - non - ab ] -.1em.8ex-.6emdy^= a^_j d_j which is an equation that is fully determined by the controls
. the price one pays is that the @xmath67 coordinates can not be simply added to calculate the total change in a stroke @xmath31 , since one has to consider the changes in the reference frame as well . in order to do that
, @xmath67 must be viewed as ( infinitesimal ) elements of the euclidean group [ eq : euclidean ] e(y^)= ( ccc y^0 & y^0 & y^1 + - y^0 & y^0 & y^2 + 0 & 0 & 1 + ) the composition of @xmath67 along a stroke @xmath31 is a matrix multiplication [ eq : matrix - prod ] e()= _ e(dy^ ( ) ) the product is , of course , non commutative .
we denote generators of translations and rotations by [ eq : translations ] e^0= ( ccc 0 & 1 & 0 + -1 & 0 & 0 + 0 & 0 & 0 + ) , e^1= ( ccc 0 & 0 & 1 + 0 & 0 & 0 + 0 & 0 & 0 + ) , e^2= ( ccc 0 & 0 & 0 + 0 & 0 & 1 + 0 & 0 & 0 + ) they satisfy the lie algebra [ e^0,e^]=-^0e^,=0 where @xmath68 is the completely anti - symmetric tensor .
one can write eq .
( [ eq : linear - response - non - ab ] ) concisely as a matrix equation [ eq : matrix - form ] -.1em.8ex-.6emdy = a_j d_j,-.1em.8ex-.6emdy= y^e^ , a_j = a_j^e^ where @xmath67 and @xmath69 are @xmath70 matrices ( summation over repeated indices implied ) . ,
plotted with the flat measure on @xmath71,title="fig:",width=453 ] + plotted using the measure induced by dissipation , title="fig:",width=377 ] + once the ( six ) transport coefficients @xmath72 are known , one can , in principle , simply integrate the system of three , first order , non - linear ordinary differential equations , eq .
( [ eq : linear - response ] ) . this can normally be done only numerically .
numerical integration is practical and useful , but not directly insightful .
we want to describe tools that allow for a qualitative understating swimming in the plane without actually solving any differential equation .
low reynolds numbers swimmers perform lots of mutually cancelling maneuvers with a small net effect .
the swimming curvature measure only what fails to cancel for infinitesimal strokes . since reversing
a loop reverses the response , it is natural to expect , that @xmath73 for a closed ( square ) loop is proportional to the area form . integrating eq .
( [ eq : matrix - form ] ) around a closed infinitesimal loop gives y = d_1d_2 where = _ 1 a_2- _ 2 a_1 -[a_1,a_2 ] , = ^e^,_j= @xmath30 and @xmath74 are @xmath70 matrices .
@xmath30 has the structure of curvature of a non - abelian gauge field @xcite . in coordinates
, this reads ^= _ 1 a^_2- _ 2 a^_1 + ^0
( a^0_1 a^_2 - a^0_2 a^_1 ) , in the lab coordinates one has , of course , x^=^d_1d_2 , ^= r^ ( ) ^ ( ) , the curvature is abelian when the commutator vanishes .
this is the case in eq .
( [ eq : numerical - curvature ] ) and it is also the case for the rotational curvature .
the abelian curvature gives full information on the swimming of finite stroke by simple application of stokes formula .
this is , unfortunately , not the case in the non - abelian case .
one can not reconstruct the translational motion of a large stroke from the infinitesimal closed strokes @xmath74 because stokes theorem only works for commutative coordinates and @xmath74 are not .
+ for purcell swimmer , @xmath30 although explicit , is rather complicated . since the dissipation metric is complicated too , we give two plots of @xmath30 : figs .
( [ fig : fphieta=2 ] , [ fig : purcellcurvx],[fig : purcellcurvy ] ) give the curvature relative to the flat measure on @xmath71 , and describe how far the swimmer swims for small strokes . in figs .
( [ fig : fphinorm ] , [ fig : purcellcurvxnorm],[fig : purcellfynorm ] ) the curvature is plotted relative to the dissipation measure and it displays the energy efficiency of strokes . using cox theory , in a manner analogous to what was done in for the symmetric swimmer , one can calculate explicitly the force ( and torque ) on the @xmath75-th rod in the form f_a^= f_aj^_j + f_a^ x^ where @xmath76 are explicit and relatively simple functions of the controls ( compare with eq .
( [ eq : f_jk ] ) ) . the swimming equation are then given by _
a f_a^= ( _ af_aj^)_j + ( _ a f_a^ ) x^=0 this reduces the problem of finding the connections @xmath77 to a problem in linear algebra .
formally ^_j=(_a f_a^)^-1 ( _ af_aj^ ) where the bracket on the left is interpreted as a @xmath70 matrix , with entries @xmath78 , and the inverse means an inverse in the sense of matrices .
although this is an inverse of only a @xmath79 matrix the resulting expressions are not very insightful .
we spare the reader this ugliness which is best done using a computer program . picking the center point of the body as the reference fiducial point is , in the terminology of wilczek and shapere @xcite a choice of gauge .
this particular choice is nice because it implies symmetries of the connection @xmath77 @xcite .
observe first that the interchange @xmath80 corresponds to a rotation of the swimmer by @xmath81 . plugging this in eq .
( [ eq : ra ] ) one finds [ eq : symmetries - r ] a^_1 ( _ 1,_2)=\ { ll + a^_2 ( -_2,-_1 ) , & + -a^_2 ( -_2,-_1 ) , & + .
this relates the two half of the square divided by the diagonal @xmath82 .
a second symmetry comes from the interchange @xmath83 corresponding to the reflection of the swimmer around the central vertical of the middle link . some reflection shows then that [ eq : symmetries-2 ] a^_1 ( _ 1,_2)=\ { ll + a^_2 ( _ 2,_1 ) , & + -a^_2 ( _ 2,_1 ) , & + .
this relates the two halves of the square divided by the diagonal @xmath84 .
the symmetries can be combined to yield the result that @xmath85 and @xmath86 are anti - symmetric and @xmath87 is symmetric under inversion a^0_j()=-a^0_j(- ) , a^1_j()=a^1_j(- ) , a^2_j()=-a^2_j(- ) the rotational motion of purcell swimmer , in any finite stroke , is fully captured by the abelian curvature ^0 = ^0=_1 a^0_2- _ 2 a^0_1 this reflects the fact that rotations in the plane are commutative . the symmetry of eq .
( [ eq : symmetries - r ] ) implies [ eq : symmetries - r - d ] ( _ 2a^0_1 ) ( _ 1,_2)= ( _ 1a^0_2 ) ( -_2,-_1 ) and this says that @xmath88is _ ani - symmetric _ under reflection in the diagonal @xmath82 .
similarly , eq . ( [ eq : symmetries-2 ] ) implies [ eq : symmetries-2 ] ( _ 2a^0_1 ) ( _ 1,_2)= -(_1a^0_2 ) ( _ 2,_1 ) , and this says that @xmath89 is _ symmetric _ about the line @xmath84 .
( [ fig : fphieta=2 ] ) is a plot of the curvature and it clearly has the requisite symmetries . the total curvature associated with the full square of shape space vanishes ( by symmetry ) . for @xmath90 , one can see in fig .
[ fig : fphieta=2 ] three positive islands surrounded by three negative lakes .
the total curvature associated with the three islands is quite small , about 0.1 .
this means that purcell swimmer with @xmath90 turns only a small fraction of a circle in any full stroke . shown with the the tam and hosoi optimal distance stroke @xcite .
the curvature is given relative to the flat measure in @xmath71 .
, title="fig:",width=529 ] + .
the curvature is given relative to the flat measure in @xmath71 .
, width=453 ] the curvatures corresponding to the two translations of a swimmer with @xmath91 are shown in figs .
( [ fig : purcellcurvx],[fig : purcellcurvy],[fig : purcellcurvxnorm],[fig : purcellfynorm])(here we use @xmath91 for comparison with @xcite ) .
the symmetries of the figures are a consequence of eqs .
( [ eq : symmetries - r],[eq : symmetries-2 ] ) .
form the first we have [ eq : symmetries-3 ] ( _ 2a^_1 ) ( _ 1,_2)=- ( _ 1a^_2 ) ( -_2,-_1 ) , = 1,2 which implies that @xmath92 are _ symmetric _ under reflection in the diagonal @xmath82 . similarly , from the eq .
( [ eq : symmetries-2 ] ) we have [ eq : symmetries-4 ] ( _ 2a^_1 ) ( _ 1,_2)=\ { ll + ( _ 1a^2_2 ) ( _ 2,_1 ) , & + -(_1a^1_2 ) ( _ 2,_1 ) , & + .
this says that @xmath93 _ symmetric _ and @xmath94 _ anti - symmetric _ under reflection in the diagonal @xmath84 .
the curvatures for the translations is non - abelian and can not be used to _ calculate _ the swimming distances for _ finite _ strokes because the stokes theorem fails .
the landscape figures for the translational curvatures provide precise information on the swimming distance for infinitesimal strokes .
they are then also useful to characterize small strokes .
the question is how small is small ? for a stroke of size @xmath52 , the controls are of size @xmath95 and the swimming distance measures by the curvature is @xmath96 .
the error in this has terms .
] of the form @xmath97 .
this suggest that the relative error in the swimming distance as measured a finite stroke is of the order @xmath98 .
hence , a stroke is small provided @xmath99 .
clearly , a purcell swimmer swims substantially less than an arm length as the arm moves .
this says that @xmath100 and so strokes of the order of a radian can be viewed as small strokes .
the x - curvature is symmetric under inversion ^1()= ^1(- ) since both @xmath85 and @xmath86 are antisymmetric under inversion , one sees that the non - abelian part of the x - curvature is of order @xmath101 near the origin . the x - translational curvature , which is non - zero near the origin , is almost abelian for small strokes .
the y - curvature , in contrast , is anti - symmetric under inversion ^12()= -^2(- ) and so vanishes linearly at the origin .
the non - abelian part is also anti - symmetric under inversion , and it too vanishes linearly .
the y - curvature is therefore _ not _ approximately abelian for small strokes , but it is small . the swimming direction can be easily determined for those strokes that live in a region where the translational curvature has a fixed sign . this answers purcell s question for many strokes .
strokes the enclose both signs of the curvature are subtle .
purcell s swimmer can reverse its direction of propagation by increasing the stroke amplitude @xcite .
this can be seen from the landscape diagram , fig .
( [ fig : purcellcurvx ] ) : small square strokes near the origin sample only slightly negative curvature .
as the stroke amplitude increases the square gets larger and begins to sample regions where the curvature has the opposite sign , eventually sampling regions with substantial positive curvature .
the curvature landscapes are useful when one wants to search for optimal strokes as they provide an initial guess for the stroke .
( this initial guess can then be improved by standard optimization numerical methods . ) for example , tam and hosi @xcite looked for strokes that cover the largest possible distance . for strokes near the origin , a local optimizer is the stroke that bounds the approximate square blue region in fig .
[ fig : purcellcurvx ] ( in this case @xmath91 ) . the curvature normalized by dissipation , fig .
[ fig : purcellcurvxnorm ] gives a guide for finding efficient small strokes .
caution must be made , since while the displacement can be approximated from the surface area , the energy dissipation is proportional to the stroke s length and not the stroke s area . in regimes
where the gaussian curvature of the dissipation ( fig .
[ fig : gausscurvpurcell ] ) is positive - it is possible to have strokes with small length which bounds large area . in the case of purcell
s swimmer , this suggest two possible regimes : around the origin and the positive curvature island in the upper left ( lower right ) corner of fig . [
fig : gausscurvpurcell ] .
the optimizer near the origin is the optimal stroke found in @xcite , while the optimizer in the upper left corner @xcite - although more efficient ( pay attention to the values of @xmath102 in fig . [ fig : purcellfynorm ] ) , is of mathematical interest only , since it is near the boundary , where the first order slender body approximation eq .
( [ eq : cox ] ) is relevant only for extremely slender bodies . | we develop a qualitative geometric approach to swimming at low reynolds number which avoids solving differential equations and uses instead landscape figures of two notions of curvatures : the swimming curvature and the curvature derived from dissipation .
this approach gives complete information for swimmers that swim on a line without rotations and gives the main qualitative features for general swimmers that can also rotate .
we illustrate this approach for a symmetric version of purcell s swimmer which we solve by elementary analytical means within slender body theory .
we then apply the theory to derive the basic qualitative properties of purcell s swimmer . |
many observed spectral features of active galactic nuclei ( agn ) indicate that outflows are common in these systems . the prominent broad emission lines ( bel ) in ultraviolet ( uv ) from h
i , o vi , n v , c iv , and si iv are the defining feature of quasars ( blandford , netzer & woltjer 1990 ; osterbrock 1989 ) , and they may be associated with a high velocity outflow .
fast outflows can also explain narrow uv absorption lines from highly ionized species such as c iv and n v observed in approximately half of the hst - observed seyfert 1 galaxies ( e.g. , crenshaw 1997 ) .
these narrow absorption lines are all blueshifted relative to the systemic velocity by 0 to -1600 km s@xmath10 .
recent asca observations show that in a small sample observed by hst and asca , all of the seyfert galaxies with warm absorbers also show intrinsic uv absorption .
a high - resolution x - ray observation of the seyfert galaxy ngc 5548 obtained by chandra shows strong , narrow absorption lines from highly ionized species ( kaastra et al .
the lines are blueshifted by a few hundred km @xmath11 and they are reminiscent of the narrow absorption lines observed in the uv .
some qsos also have intrinsic uv narrow - line absorbers with line - of - sight velocities as large as @xmath12 ( hamann et al .
perhaps the most compelling evidence for fast outflows in qsos are strong broad absorption lines ( bals ) in the uv resonance lines of highly ionized species such as n v , c iv , and si iv .
the bals are always blueshifted relative to the emission - line rest frame , indicating the presence of outflows from the active nucleus , with velocities as large as @xmath13 .
a key constraint on any model for the origin of agn outflows is the ionization balance .
on one hand we observe very high luminosities in x - rays and the uv and on the other hand we observe spectral lines from moderately and highly ionized species .
one wonders then how the gas avoids full photoionization and we see any spectral lines at all .
generally , two mechanisms have been proposed to resolve the so - called overionizaton problem : ( i ) the agn outflows have filling factors less than one and consist of dense clouds and ( ii ) the filling factors equal one but the outflows are shielded from the powerful radiation by some material located between the central engine and the outflow ( e.g. , krolik 1999 , and below ) .
many theoretical models have been proposed to explain outflows in agns .
comprehensive reviews of recent theoretical work on outflows in agns can be found in arav , shlosman & weymann ( 1997 ) .
several forces have been suggested to accelerate outflows in agn , for example , gas pressure ( e.g. , weymann et al .
1982 ; begelman , de kool & sikora 1991 ) , magneto - centrifugal force due to an accretion disk ( e.g. , blandford & payne 1982 ; emmering , blandford & shlosman 1992 ; k@xmath14nigl & kartje 1994 ; bottorff et al .
1997 ) , radiation pressure due to dust ( e.g. , voit , weymann & korista 1993 ; scoville & norman 1995 ) , and radiation pressure due to lines ( e.g. , drew & boksenberg 1984 ; shlosman , vitello & shaviv 1985 ; arav , li & begelman 1994 ; de kool & begelman 1995 , hereafter dkb ; murray et al . 1995 , hereafter mcgv ) .
most of these models avoid overionization of the wind by assuming high density clouds .
one plausible scenario for the agn outflows is that radiation pressure on spectral lines drives a wind from an accretion disk around a black hole .
the presence of the bals themselves strongly indicates that substantial momentum is transfered from a powerful radiation field to the gas .
a crucial clue to the origin of agn outflows comes from the discovery of line - locking in bal qso spectra ( e.g. , foltz et al .
weymann et al .
( 1991 ) discovered that a composite spectrum of their bal qso sample shows a double trough in the c iv @xmath151549 bal , separated in velocity space by the n v @xmath15 1240-ly@xmath0 splitting @xmath16 km @xmath11 ( see also korista et al .
in fact arav & begelman ( 1994 , see also arav et al .
1995 and arav 1996 ) showed that an absorption hump in the c iv @xmath15 bal is `` the ghost of ly@xmath0 '' due to modulation of the radiation force by the strong emission in ly@xmath0 .
our understanding of how line - driving produces powerful high velocity winds is based on the studies of winds in hot stars that radiate mostly in the uv ( e.g. , castor , abbott , klein 1975 , hereafter cak ; abbott 1982 ) .
the essential concept underpinning these models is that the momentum is extracted most efficiently from the radiation field via line opacity . with the inclusion of lines
, cak showed that the effective radiation force can increase by several orders of magnitude above that due to electron - scattering alone .
thus even a star that radiates at around 0.1% of its eddington limit , @xmath17 can have a strong wind . up to now , however , it has been difficult to apply line - driven stellar wind models to qsos because of the fundamental difference in geometry : stellar winds are to a good approximation spherically symmetric , whereas the wind in bal qsos likely arises from a flattened disk and is therefore axially symmetric .
line - radiation driven disk wind models for bal qsos have been proposed by several authors .
for example , mcgv studied a wind arising from a small distance from the central engine ( @xmath3 cm ) .
their model relies on the local disk radiation to launch gas from the disk and on the central radiation to accelerate the gas in the radial direction to high velocities .
a key ingredient to this model is that the central engine radiation is attenuated by hitchhiking gas located between the central engine and wind . rather than showing from first principles the origin of this hitchhiking gas
, they simply give plausible arguments for its existence .
mcgv adopted assumptions that require the flow to be time - independent and restrict the flow geometry to a quasi 1-d radial flow .
other models of radiation - driven outflows in agns propose that the radiation force accelerates the wind that has been launched and kept at a low ionization state by _ different _ mechanisms .
for example , the dkb model supposes that the wind is seated at a large distance ( @xmath18 cm ) from the central engine . because the disk is cool at these radii , the gas is not lifted from the disk by radiation pressure but by some alternative mechanism .
once the gas is high enough above the disk , the radiation force due to the central engine accelerates the gas to high velocity .
the gas is not over - ionized in this model due to a small filling factor caused by strong magnetic fields which confine the dense clouds .
recently it has been possible to model line - driven disk winds using 2-dimensional axisymmetric numerical hydrodynamical simulations ( e.g. , pereyra , kallman & blondin 1997 ; psd i ; proga 1999 ; proga , stone & drew 1999 , hereafter psd ii ) .
all these models have been calculated for white dwarf accretion disks .
psd i found that line - driven disk winds are produced only when the effective luminosity of the disk ( i.e. , the luminosity of the disk times the maximum value of the force multiplier , @xmath19 ) exceeds the eddington limit , @xmath17 .
if the dominant contribution to the total radiation field comes from the disk , then the outflow is intrinsically unsteady and characterized by large amplitude velocity and density fluctuations . on the other hand ,
if the total luminosity of the system is dominated by the central object , then the outflow is steady . in either case
, psd i and psd ii found that the structure of the wind consists of a dense , slow outflow , that is bounded on the polar side by a high - velocity , lower density stream .
the flow geometry is controlled largely by the geometry of the radiation field
a brighter disk / central object produces a more polar / equatorial wind .
global properties such as the total mass loss rate and terminal velocity depend more on the system luminosity and are insensitive to geometry .
the mass loss rate is a strong function of the effective eddington luminosity and is of the same order of magnitude as that of a simple spherically - symmetric stellar wind ( see also proga 1999 ) . matter is fed into the fast stream from within a few central object radii .
the terminal velocity of the stream is similar to that of the terminal velocity of a corresponding spherical stellar wind , i.e. , @xmath20 , where @xmath21 , is the escape velocity from the photosphere . in this paper
, we use the methods developed by psd to study disk winds in agns .
however to study agns winds , we need to relax some of the assumptions adopted by psd .
in particular , we can no longer assume the outflowing gas is isothermal , has a fixed ionization state , or is optically thin .
instead , we adopt a simplified treatment of photoionization , and radiative cooling and heating that allow us to compute self - consistently the ionization state , and therefore line force , in the wind . here
we calculate a few disk wind models using our extensions to the psd ii method .
we concentrate on assessing how winds can be driven from a disk in the presence of very strong ionizing radiation .
we describe our calculation in section 2 .
we present our results in section 3 and discuss them together with perceived limitations in section 4 .
the paper ends , in section 5 , with our conclusions .
to calculate the structure and evolution of a wind from a disk , we solve the equations of hydrodynamics @xmath22
@xmath23 @xmath24 where @xmath25 is the mass density , @xmath26 is the gas pressure , @xmath27 is the velocity , @xmath28 is the internal energy density , @xmath29 is the net cooling rate , @xmath30 is the gravitational acceleration of the central object , @xmath31 is the total radiation force per unit mass .
we adopt an adiabatic equation of state @xmath32 , and consider models with @xmath33 .
our calculations are performed in spherical polar coordinates @xmath34 .
we assume axial symmetry about the rotational axis of the accretion disk ( @xmath35 ) .
however the @xmath36 axis is not coincident with the disk midplane .
we allow this axis to be above the midplane , at the height , @xmath37 , corresponding to the disk pressure scale height .
figure 1 shows a schematic representation of our computation domain .
the offset enters our calculations in the computation of the gravity ; for example the gravity has non - zero @xmath38 component at @xmath39 , and in the computation of the keplerian velocity along the @xmath40 axis .
this offset is introduced so that the bottom of our computational grid is located at the disk photosphere , where the wind is launched . unlike the thin disk studied by psd , the radiation dominated disk in agn may have significant geometrical width .
if the base of the computational grid is placed at the equatorial plane ( midplane ) , this would require modeling the full internal structure of the disk , which is complex and turbulent ( e.g. , balbus & hawley 1998 ) .
instead , we locate the base of the grid where the vertical radiation force due to electron scattering cancels out the vertical component of gravity .
a realistic description of the radiation field from the central engine would require detailed knowledge of the geometry of the flow near the central engine , which is beyond the scope of this investigation .
instead we model the central engine as a point source of radiation located at the origin of our computational grid ( that is located a height @xmath37 above the disk midplane ) .
this implies the central engine has finite width , perhaps associated with a hot corona .
we expect the high column density of the disk to attenuate radiation close to the @xmath39 plane .
our standard computational domain is defined to occupy the radial range @xmath41 , where @xmath42 is the inner radius of the disk , and the angular range @xmath43 . for comparison
, we also calculate some models with the radial range @xmath44 and with the radial range @xmath45 .
the @xmath46 domain is discretized into zones .
our numerical resolution consists of 100 zones in each of the @xmath47 and @xmath38 directions , with fixed zone size ratios , @xmath48 and @xmath49 .
gridding in this manner ensures good spatial resolution close to the radiating surface of the disk and the inner boundary at @xmath50 .
the boundary conditions are specified as follows . at @xmath51
, we apply an axis - of - symmetry boundary condition . for the outer radial boundary
, we apply an outflow boundary condition . for the inner radial boundary @xmath52 and for @xmath39 , we apply reflecting boundary conditions for the density , velocity and internal energy . to represent steady conditions in the photosphere at the base of the wind , during the evolution of each model
we apply the constraints that in the first zone above the @xmath39 plane the radial velocity @xmath53 , the rotational velocity @xmath54 remains keplerian , and the density is fixed at @xmath55 at all times .
the initial conditions are as in psd i expect for the initial temperature profile which is described in detail in section 2.3 . to solve eqs .
1 - 3 we use an extended version of the zeus-2d code ( e.g. , stone & norman 1992 ) .
the equation for the internal energy , eq .
3 , is solved using the operator splitting method and the backward euler scheme . the geometry and assumptions needed to compute the radiation field from the disk and central object are as in psd ii ( see also psd i ) .
the disk is flat , keplerian , geometrically - thin and optically - thick .
the disk photosphere is coincident with the @xmath39 axis .
we specify the radiation field of the disk by assuming that the temperature follows the radial profile of the optically thick accretion disk ( shakura & sunyaev 1973 ) , and therefore depends on the mass accretion rate in the disk , @xmath56 , the mass of the black hole , @xmath57 and the inner edge of the disk , @xmath58 , where @xmath59 is the schwarzschild radius of a black hole . in particular , the disk luminosity , @xmath60 , where @xmath61 is the rest mass conversion efficiency .
the geometry of the central engine in agns is not well known . for simplicity , we consider the central engine as the most inner part of the accretion disk plus an extended corona .
we refer to the corona as the central object .
the radius of the central object is comparable with the inner radius of the disk and we assume that they formally are equal .
we express the central object luminosity @xmath62 in units of the disk luminosity @xmath63 .
in contrast to psd i and psd ii , we allow for the situation when only some fraction of the central object luminosity takes part in driving a wind .
we identify this fraction as the luminosity in the uv band , @xmath64 .
we refer to the fraction of the luminosity that is responsible for ionizing the wind to a very high state as the luminosity in the x - ray band , @xmath65 . for simplicity , we assume here that this fraction of the luminosity does not contribute to line driving of the wind .
we call the luminosity in the remaining bands , mainly optical and infrared , as @xmath66 .
we assume that @xmath67 is the part of the luminosity that does not change the dynamics of the wind .
we stress that the fraction @xmath68 is a numerical factor that we introduce here to parameterize the luminosity in each of those three domains .
we set @xmath69 to zero in the remaining part of the paper as the central object is very hot .
we take into account the irradiation of the disk by the central object , assuming that the disk re - emits all absorbed energy locally and isotropically .
however we note that the contribution from irradiation is negligible for @xmath70 and large radii ( see eq . 6 below ) .
we approximate the radiative acceleration due to lines ( line force , for short ) using a modified cak method .
the line force at a point defined by the position vector @xmath71 is @xmath72 where @xmath73 is the frequency - integrated continuum intensity in the direction defined by the unit vector @xmath74 , and @xmath75 is the solid angle subtended by the disk and central object at the point .
the term in brackets is the electron - scattering radiation force , @xmath76 is the mass - scattering coefficient for free electrons , and @xmath77 is the force multiplier the numerical factor which parameterizes by how much spectral lines increase the scattering coefficient . in the sobolev approximation
, @xmath77 is a function of the optical depth parameter @xmath78 where @xmath79 is the thermal velocity , and @xmath80 is the velocity gradient along the line of sight , @xmath74 .
we evaluate the radiation force in four steps .
first , we calculate the intensity , the velocity gradient in the @xmath74 direction and then the optical depth parameter @xmath81 .
second , we calculate the parameters of the force multiplier using a current value of the photoionization parameter , @xmath82 adopting results of stevens & kallman ( 1990 ) . then we calculate the radiation force exerted by radiation along @xmath74 .
third , we integrate the radiation force over the solid angle subtended by the radiant surface .
finally , we correct the radiation force in the radial direction for the optical depth effects . our numerical algorithm for evaluating the line force for given parameters of the force multiplier , the third step , is described in psd ii . for a rotating flow
, there may be an azimuthal component to the line force even in axisymmetry .
however we set this component of the line force to zero because it is always less than other components ( e.g. , psd ii ) .
see psd i and psd ii for further details .
below we describe our calculations of the force multiplier for various conditions in the wind and our treatment of the optical depth effects on the radiation force . in the disk plane at @xmath83 , @xmath84 is the local isotropic disk intensity : @xmath85 we consider only the hot part of the disk , where the local disk temperature , @xmath86 a few @xmath87 k. we then assume that all disk photons can contribute to the line force .
the second term in the curly brackets corresponds to the contribution from the irradiation of a disk by a central object . for large radii and @xmath88 ,
the contribution from the irradiation is negligible compared to the intrinsic disk intensity ( the first term in the curly brackets ) . for radiation from the central object
, the intensity may be written as : @xmath89 we account for the fact that the central radiation consists of the two distinct spectral components by using the parameter @xmath90 , where @xmath91 or @xmath92 .
we assume that the optical depth effects in the @xmath38 direction are negligible for the central radiation in the two bands .
note that the @xmath38 component of the radiation force due to the central object is negligible in our computational domain because the domain is far from the object ( @xmath93 ) . in the radial direction
however , the column density can be high and the optical depth effects can be significant .
we estimate the optical depth , @xmath94 between the central source and a point in a wind from : @xmath95 where @xmath96 is the absorption coefficient representative for the @xmath97 band , and @xmath47 is the distance from the central source .
we can treat the central object as a point source because our calculations are for @xmath98 .
then we calculate the radial component of the central radiation force due to lines from @xmath99 where @xmath100 is the central object solid angle .
we also take into account the optical depth effects on the radial component of the electron - scattering force : @xmath101 the attenuation of the x - ray radiation is calculated using @xmath102 for @xmath103 and @xmath104 for @xmath105 , while the attenuation of the uv radiation is calculated using @xmath106 for all @xmath82
. a disk wind may also be optically thick to the uv radiation emitted by the disk .
in particular , the flux along the radial direction near the disk can be significantly reduced due to high column density .
we approximate this effect by multiplying the radial component of the disk radiation force by the attenuation factor @xmath107 .
to calculate the force multiplier , we adopt the cak analytical expression modified by owocki , castor & rybicki ( 1988 , see also psd i ) @xmath108\ ] ] where @xmath109 is proportional to the total number of lines , @xmath0 is the ratio of optically - thick to optically - thin lines , @xmath110 and @xmath111 is a parameter determining the maximum value , @xmath112 achieved for the force multiplier . equation 11 shows the following limiting behavior : @xmath113 where @xmath114 . the maximum value of the force multiplier is a function of physical parameters of the wind and radiation field .
several studies have showed that @xmath112 is roughly a few thousand for gas ionized by a weak or moderate radiation field ( e.g. , cak ; abbott 1982 ; stevens & kallman 1990 ; gayley 1995 ) . as the radiation field becomes stronger and the gas becomes more ionized the force multiplier decreases asymptotically to zero .
the line force , in particular the parameters of the force multiplier depend on the ionization of the wind and the spectral energy distribution ( sed ) of the radiation field .
self - consistent calculations of the line - force for given wind conditions and the radiation require detailed calculations of the wind photoionization structure ( e.g. , cak ; abbott 1982 ; puls et al . 2000 ) .
the total line force includes contributions from @xmath115 lines from many species .
such calculations in connection with the 2-d , time - dependent hydrodynamic calculations are not feasible .
we therefore start by adopting the analytical formulae for the force multiplier due to stevens & kallman ( 1991 ) .
they studied the effects of x - ray ionization on the the radiative force experienced by the stellar wind in a massive x - ray binary ( mxrb ) .
wind conditions in mxrbs differ somewhat from those in qsos .
in particular , the x - ray radiation in mxrbs can be well represented by a 10 kev bremsstrahlung spectrum while the spectrum in qsos is better fitted by power laws with different spectral indices at different spectral bands ( e.g. , zheng et al .
1997 ; laor et al .
nevertheless we adopt the stevens & kallman results as a first order approximation because the compton temperature and the ionization structure are similar ( kallman & mccray 1982 ) . using the photoionization code xstar ,
stevens & kallman found that the line force due to the radiation from the primary can be parameterized in terms of the cak force multiplier .
their results show that the line force decreases sharply with increasing photoionization parameter : @xmath116 where @xmath117 is the local x - ray flux , @xmath118 is the number density of the gas ( @xmath119 , where @xmath120 is the proton mass , and @xmath121 is the mean molecular weight ) .
stevens & kallman also found some simple analytical fits to their results that allowed them to express @xmath109 and @xmath111 as functions of @xmath82 : @xmath122 and @xmath123 the parameter @xmath124 and does not change with @xmath82 .
these expressions for the parameters of the force multiplier predict that @xmath112 increases gradually from @xmath125 to @xmath126 as @xmath82 increases from 0 to @xmath127 and then drops to @xmath128 at @xmath129 .
the line force becomes negligible for @xmath130 because then @xmath131 . to calculate the photoionization parameter
, we first need to know the local x - ray flux @xmath117 ( see eq .
we estimate @xmath117 assuming that the source of all x - rays is a point - like central object with the luminosity in the x - ray band , @xmath132 . making the above simplifications
, we can express the local x - ray flux as @xmath133 we calculate the gas temperature assuming that the gas is optically thin to its own cooling radiation and that the abundances are cosmic ( withbroe 1971 ) .
the net cooling rate depends on the density , @xmath25 , the temperature , @xmath134 , the ionization parameter @xmath82 , and the characteristic temperature of the x - ray radiation @xmath135 .
it is then possible to fit analytical formulae to the heating and cooling rate obtained from detailed photoionization calculations for various @xmath136 , @xmath134 , and @xmath82 .
for example , blondin ( 1994 ) found that for a 10 kev bremsstrahlung spectrum , his fits typical agree with a detailed calculation ( blondin et al .
1990 ) to within 25% . using blondin s result
we can express the net cooling rate @xmath137 in equation 3 by @xmath138 where @xmath139 is the rate of compton heating / cooling , @xmath140 @xmath141 is the net rate of x - ray photoionization heating
recombination cooling @xmath142 and @xmath143 is the rate of bremsstrahlung and line cooling @xmath144 \delta ~\rm~erg~\rm cm^{-3}~s^{-1}.\ ] ] the parameter @xmath145 is introduced to control line cooling , @xmath146 represents optically thin cooling and @xmath147 represents the case when lines become optically thick and cooling is reduced . as we mentioned in section 2.1
, we use the same initial and boundary conditions as in psd i. however we here also calculate the evolution of the internal energy .
we specify the boundary conditions and the initial conditions for @xmath28 as follows . for the temperature at the base of the wind , in the first grid zone above the @xmath39 plane
, we adopt the solution for the steady state disk . adopting the assumption that the local disk intensity is a black body and using eq .
6 , the disk temperature is : @xmath148 then we calculate the internal energy from @xmath149 where @xmath150 is the density in the first grid zone above the @xmath39 plane .
we consider models with @xmath151 .
we initialize the internal energy at other locations by assuming that the gas temperature at a given radius is the same as at the disk temperature ( i.e. , @xmath152 for all @xmath38 ) and using initial value of the density ( see psd i ) .
consistent with the boundary conditions for other gas quantities ( see sect .
2.1 ) we apply the constraint that in the first zone above the @xmath39 plane the internal energy @xmath153 at all times .
we find only the initial , transient evolution of the wind is affected by this choice of initial temperature conditions .
our treatment of radiative heating and cooling is most likely valid in a low density and high photoionization parameter regime . for a high density and low photoionization parameter regime our treatment will likely underestimate the temperature because we oversimplify line cooling by not properly taking into account optical depth effects . in dense regions ,
our scheme therefore may yield a very low gas temperature ( e.g. , @xmath154 k ) .
in such cases we set the lower limit for the temperature assuming that the gas is in local thermodynamical equilibrium , i.e. , the gas temperature equals the local disk radiation temperature .
specifically , if at the ( @xmath155 ) wind point the temperature from eq . 3 is less than @xmath156 then we replace it with @xmath156 .
we specify our models by several parameters . in all our calculations
we assume the mass of the non - rotating black hole , @xmath157 and the rest mass conversion efficiency , @xmath158 typical for agn . to determine the radiation field from the disk , we assume the mass accretion rate @xmath159 m@xmath160 yr@xmath10 . these system parameters yield the disk eddington number , @xmath161 , the disk inner radius , @xmath162 cm and the orbital period at the inner edge of the disk , @xmath163 sec .
we set the offset of the computational domain , @xmath164 that corresponds to the half width of the radiation pressure dominated disk at large radii ( e.g. , shakura & sunyaev 1973 ) .
the radiation field from the central engine is specified by the additional parameters : @xmath165 , @xmath166 , @xmath167 . as we discussed above the sed of the ionizing radiation
is not well known , our choice of values for @xmath168 and @xmath169 is guided by the results from zheng et al .
( 1997 ) and laor et al .
( 1997 ) . to calculate the line force
we adopt the force multiplier parameter @xmath124 .
we calculate the remaining parameters of the multiplier , i.e. , @xmath109 and @xmath111 as functions of the photoionization parameter , @xmath82 using eqs 15 and 16 .
the force multiplier depends only formally on the thermal speed , @xmath79 which we set to 20 @xmath170 , i.e. , the thermal speed of a hydrogen atom at the temperature of 25000 k ( stevens & kallman 1990 ) . to calculate the gas temperature
, we assume the temperature of the x - ray radiation , @xmath171 kev and the line cooling parameter @xmath172 .
we adopted a generalized cak method to calculate a disk wind driven by the line force ( psd ii ) .
the cak method has been developed for ob stars with photospheres dominated by gas pressure ( the inner atmosphere is in hydrostatic equilibrium ) , and they radiate most of their energy in the uv band .
we chose the radial extent of the computational domain ( i.e. , @xmath173 ) keeping in mind that the disk atmosphere should satisfy these two conditions , however some compromises had to be done .
for example , the disk temperature at @xmath174 cm is 8200 k while at @xmath175 is 1500 k using our disk parameters .
we realize that according to the @xmath0-disk model the disk emits mostly in the uv band down to the radius of @xmath176 where its effective temperature is @xmath177 k. however the disk at this small radius is likely radiation dominated ( e.g. , shakura & sunyaev 1973 ; svensson & zdziarski 1994 ) . on the other hand ,
the disk atmosphere at radius @xmath178 and the effective temperature of @xmath179 , radiates most of its energy in the infrared band .
nevertheless we chose a large value for the outer edge of the computational domain to make sure that the domain includes most of the acceleration zone of the disk wind launched from the inner disk and the wind velocity at the outer edge well represents the wind terminal velocity .
figure 2 shows the instantaneous density , temperature and photoionization parameter distributions and the poloidal velocity field of the model .
after @xmath180 years ( @xmath181 ) , the disk material fills the grid for @xmath182 and remains in that region for the rest of the run , i.e. , over next @xmath183 years .
figure 2 shows results after @xmath184 years .
although the flow is time - dependent the gross properties of the flow ( e.g. , the mass loss rate and the radial velocity at the outer boundary ) , settle down to steady time - averages over timescales on the order of @xmath185 years . our calculation follows ( i ) a hot , low density flow in the polar region ( ii ) a dense , warm and fast equatorial outflow from the disk , ( iii ) a transitional zone in which the disk outflow is hot and struggles to escape the system . in the polar region ,
the density is very small and close to the lower limit that we set on the grid , i.e. , @xmath186 .
the line force is negligible because the matter is highly ionized as indicated by a very large photoionization parameter ( @xmath187 ) .
the gas temperature is close to the temperature of the x - ray radiation , again indicative of highly ionized gas .
the matter in the polar region is pulled by the gravity from the outer boundary and it is an artifact of the boundary conditions .
overall this region of the very low density is not relevant to our analysis as it has no effect on the much denser disk flow .
the equatorial region is distinctly different . in the inner part of the disk ( i.e. , for @xmath188 ) ,
the density at the wind base is high , @xmath189 .
thus the photoionization parameter is low despite the strong central radiation .
however as the flow from the inner part of the disk is accelerated by the line force its density decreases and the gas temperature and the photoionization parameter increase .
subsequently the gas becomes fully ionized and loses all of driving lines before it reaches the escape velocity and therefore falls on the central object / inner disk .
although this gas does not produce a wind , its primary effect is to shield the gas at larger radii .
thus the wind consists of gas accelerated by the line force at larger radii , in fact driving of the disk wind extends over all radii at which the intrinsic disk radiation is large enough to launch gas . the poloidal velocity ( fig .
2c ) shows that the gas streamlines are perpendicular to the disk over some height that increases with radius .
the streamlines then bend away from the central object and converge .
the region where the flow is moving almost radially outward is associated with a high - velocity , high density stream .
this fast stream contributes @xmath190 100% to the total mass loss rate , @xmath191 .
we note that the mass loss rate can increase by a factor of a few when a knot is crossing the outer boundary .
the fast stream is variable . in the upper envelope of the disk wind
there is large velocity shear between the higher density fast stream moving outward and the lower density fast gas moving inward .
our simulations show that this shear gives rise to kelvin - helmholtz instabilities .
the instabilities generate knots that propagate along the fast stream at @xmath192 km @xmath11 .
figure 2a shows an example of such a knot at @xmath193 and @xmath194 , in the figure coordinates .
the density contrast between the knot and the fast stream is @xmath195 orders of magnitude .
the knots generator is episodic and is inherent to the fast stream .
we observed the generation of 19 knots over @xmath196 years .
in other words , a knot is produced every @xmath127 years .
figure 3 presents a sequence of density maps showing time evolution of the outflow from figure 2 after 13.3 , 14.6 and 16.47 years , left , middle and righ panel respectively .
note that the middle panel from figure 3 is the same as the top left panel from figure 2 .
figure 3 well illustrates variability in the outflow , in particular , generation of a knot in the fast stream at @xmath197 and @xmath198 ( left panel ) , the well - formed knot at @xmath193 and @xmath194 ( middle panel ) , and the time when the knot left the gird at @xmath199 and @xmath200 , and the fast stream is fairly smooth in between episodes with knots ( right panel ) .
the time dependence in the region close to the disk , bounded by the fast stream resembles outward propagation of a wave at the begin of evolution but becomes more complex with time as elements of the flow move both upwards and downwards .
figure 4 presents the run of the density , radial velocity , photoionization parameter and column density as a function of the polar angle , @xmath38 at the outer boundary , @xmath201 cm from figure 2 .
the column density is given by : @xmath202 the gas density is a very strong function of angle for @xmath38 between @xmath203 and @xmath204
. the density drops by @xmath205 orders of magnitude between @xmath206 and @xmath207 , as might be expected of a density profile determined by hydrostatic equilibrium . for @xmath208 , the wind domain , @xmath25 varies between @xmath209 and @xmath210 . for @xmath211 ,
density decreases gradually to so low a value that it becomes necessary to replace it by the numerical lower limit @xmath212 .
the radial velocity at @xmath213 cm has a broad peak for @xmath214 with the maximum of 15000 km @xmath11 at @xmath215 . on both side of the peak
the velocity is close to zero .
the photoionization parameter is very high , @xmath187 , for @xmath211 because of the very low density .
however it drops by 15 orders of magnitude between @xmath216 and stays at a very low level for @xmath217 .
the column density changes less dramatically with angle . between the pole and @xmath218
, @xmath219 is less than @xmath220 . for @xmath221
, @xmath219 increases gradually with @xmath38 , it reaches value of @xmath222 at @xmath223 . for @xmath224
, @xmath219 varies between @xmath222 and @xmath225 .
our model shows that ( i ) the intrinsic disk radiation can launch a wind and ( ii ) the central object radiation can accelerate the disk wind to very high velocities .
we have checked how these main results are sensitive to the model parameters . for a fixed disk atmosphere and central radiation source , the most important parameter of our model is the optical depth for the x - rays from the central object .
therefore we have run several tests with various x - ray opacities for @xmath226 : @xmath227 and @xmath228 .
the former case is most conservative because it corresponds to the case when the x - ray optical depth is reduced to the thomson optical depth .
we found that in both test runs the x - ray radiation is significantly attenuated so the line force can launch a disk outflow .
however the flow velocity never exceeds the escape velocity because the x - rays fully ionize the gas close to the disk and produces a hot corona with complex velocity field but which does not escape the system .
we conclude then that the disk atmosphere can shield itself at least to the extent that the local disk radiation can launch gas off the disk photosphere .
the two test runs also illustrate how robust is our second result : the central radiation accelerates the wind to very high velocities . for this to happen the column density
, @xmath229 must be large enough to reduce the x - ray radiation but too not large to reduce the uv flux in the radial direction .
in other words , this requires @xmath230 and at the same time , @xmath231 .
the test run with @xmath232 corresponds to the situation where @xmath233 .
thus the disk gas that is sufficiently shielded from the x - rays is also shielded from the uv photons from the central object .
such gas can only be lifted from the disk surface by the disk uv radiation but fails to gain momentum in the radial direction .
the fact that the disk wind can be launched without _
external _ shielding material implies that our solution can depend on the location of the inner computational radius , @xmath50 .
we have run several tests with @xmath234 and @xmath235 to check this ( the remaining model parameters were as in the models shown in figure 1 and 2 ) .
we found that the location of the inner edge , @xmath50 affects the properties of the wind but not the fact the wind is produced .
for example , the wind opening angle , @xmath236 , is @xmath237 , @xmath238 and @xmath239 for @xmath240 and @xmath241 , respectively . the maximum radial velocity at the outer boundary decreases from 20000 @xmath5 through , 15000 @xmath5 to 5000 @xmath5 for above sequence of decreasing @xmath50 .
the mass loss rate also decreases with @xmath50 from 0.6 @xmath242 through 0.5 @xmath242 to 0.2 @xmath242 .
our calculations show that the uv emitting accretion disk can launch a wind that will shield itself from the strong ionizing radiation emitted from the central object .
the self - shielding is quite robust because the external radiation does not penetrate the disk down to the region where the line force becomes important and pushes matter away from the photosphere .
we illustrated the robustness of the self - shielding by changing relevant model parameters .
we also note that a similar effect was observed by stevens & kallman ( 1990 ) and stevens ( 1991 ) .
they studied the effects of x - ray ionization on the radiative force experienced by the stellar wind in a mxrb .
they calculated the detailed photoionization structure of the line - driven wind for various x - ray luminosities without optical depth effects ( stevens & kallman 1990 ) and with optical depth effects ( stevens 1991 ) .
additionally , they considered irradiation at various directions including the normal to the mass losing photosphere . in none of the models the irradiation penetrated below the so - called wind critical point where , in the cak type models , the wind mass loss rate is determined .
they found , as we did , that the irradiation can severely decrease the wind velocity in the supersonic portion of the flow .
we model here a line - driven wind from a disk that is flat , keplerian , geometrically - thin and optically - thick .
for the temperature at the base of the wind , in the first grid zone above the @xmath39 plane , we adopt the solution for the steady state @xmath0 disk .
we calculate the local disk intensity as if it emits as a black body . in all our calculations ,
we treat the regions for @xmath243 as if they are gas pressure dominated .
we find that our results depend on the inner radius of the computational domain .
this dependence is likely an artifact of one or more of our assumptions .
for example , for very high luminosities the inner part of the disk is dominated by the radiation pressure . for the parameters adopted here , namely @xmath244 , the inner radius of a gas pressure dominated disk is @xmath3 cm ( e.g. , svensson & zdziarski 1994 ) .
the structure of the radiation dominated disks is not well known .
we expect , however , that incorporating the radiation dominated part of the disk into our calculation will change the condition of the base of the wind , for example , the gas density and the uv flux might be reduced .
these changes might affect the solution for the wind , in particular , the wind mass flux .
we plan to explore these issues in a future paper .
to calculate the disk surface temperature and intensity , we took into account the irradiation of the disk by the central object .
however the contribution from the irradiation is negligible compared to the disk intrinsic temperature because we considered a flat disk at large radii .
additionally , the high column density of the disk wind , that we found in our models , implies that the irradiating flux will be significantly attenuated by the wind before it reaches the disk surface .
we anticipate that the last effect will significantly reduce disk irradiation regardless of the shape of the disk surface flat or flaring .
detailed nlte photoionization calculations are required to determine what fraction of the central radiation will reach the surface of a mass losing disk . a complete treatment of disk irradiation is difficult also because it is not obvious a priori whether the radiation incident on the disk is completely thermalized and re - radiated isotropically or it is scattered off the disk atmosphere or both .
our calculations show that a line - driven disk wind model offers a promise of explaining outflows in agns .
however they also illustrate some problems with this model .
for example , as dkb pointed out in discussing mcgv s model , a very small radius at which the disk wind is launched also implies a very small size to belr because belr lies inside or is cospatial with balr .
the small scales in turn imply short crossing time of the balr , of order of a month or so , and it is difficult to understand that the highly complex kinematical structures that bals often exhibit do not appear to vary on timescales of 10 years ( barlow 1994 ) .
our axisymmetric 2d models show that the disk wind is unsteady and generates dense knots every 3 years .
these knots will correspond to rings in 3d .
however in fully 3d calculations the rings may break down to spirals or clouds or both .
the breaking down of the rings will change the density contrast between them and the rest of the wind .
thus it is not clear if we would be able to see any spectral signature of density fluctuations in the wind .
detail calculations of line profiles are required to check this point .
additionally , detailed photoionization calculations are required to check if the full range of ions observed to show bal profiles can be explained : bals from ions with ionization potentials as low as of o iii or lower and as high as of o vi .
to produce a fast wind the ratio between @xmath245 to @xmath246 is very important .
the low ratio gives a slow disk wind whereas the high ratio gives a fast wind .
this result is consistent with the observed anti - correlation for qsos between the relative strength of the soft x - ray flux and the civ absorption equivalent width ( e.g. , brandt , laor & wills 2000 ) .
however there is an upper limit for the x - ray attenuation , or the column density between the x - ray source and wind , namely for @xmath247 ( @xmath248 ) the gas is well shielded from the x - rays but at the same time it is also shielded from any other radiation from the central engine including the uv radiation . in this case , the line force from the central engine could be so much reduced that it may not accelerate the gas to high radial velocities and no strong wind will be produced .
some of the emission - line properties of qsos could be explained as a correlation between luminosity and the slope of the ionizing spectrum , i.e. , lower luminosity objects have harder spectra ( e.g. , boroson & green 1992 ) .
using our parameterization of the agn radiation , this qso s property can be represented by increasing @xmath169 in expense of @xmath168 when we reduce the luminosity .
such a choice of model parameters will increase the strength of x - ray ionizing radiation in comparison to the uv driving radiation .
consequently a disk wind should be weaker and slower as some of our test runs indicate .
boroson & green ( 1992 ) argue that the dominant source of variation in the observed properties of low redshift qsos is not driven by external orientation but rather by the fraction of the eddington luminosity at which the object is emitting ( our parameter @xmath249 ) .
we plan to examine the parameter space of our models to define the major trends in disk wind behavior that will help us to explain the observational trends found in various agns .
we have studied radiation driven winds from luminous accretion disks using numerical methods to solve the two - dimensional , time - dependent equations of hydrodynamics . in so doing we have accounted for the radiation force due to spectral lines using a generalized multidimensional formulation of the sobolev approximation .
additionally we have taken into account the effects of the strong central radiation on the wind photoionization structure and thermodynamics .
we find that the local disk radiation can launch a wind from the disk despite strong ionizing radiation from the central object .
the central radiation may overionize the supersonic portion of the flow and severely reduce the wind velocity . to produce a fast disk wind
the wind x - ray opacity must be higher than the uv opacity by @xmath250 2 orders of magnitude .
our calculations of a wind from a disk accreting onto a @xmath251 black hole at the rate of 1.8 @xmath242 show that the radiation force from an accretion disk can launch a self - shielding wind from a radius of @xmath252 cm while the strong uv radiation from the central object can radially accelerate the disk wind to velocities @xmath253 km @xmath11 , for the x - ray opacity of 40 @xmath254 .
the disk wind domain is intrinsically unsteady and its covering factor is @xmath7 .
the wind mass loss is 0.5 @xmath242 which is a significant fraction of the mass accretion rate .
the strong x - ray radiation from a central object completely ionizes the polar region and only a thin layer above the upper envelope of the disk wind .
the disk wind immediately below the upper envelope can be characterized as a fast , high - density stream which is reminiscent of the psd disk wind solution dominated by the driving radiation from a bright central object .
the column density of the fast stream is between @xmath220 and @xmath222 so the stream is optically thin to the uv radiation and this is precisely why it can be accelerated to high velocities . on the other hand ,
the part of the wind bounded by the fast stream , closer to the disk , is slow and is reminiscent of the psd solution dominated by the driving radiation from the disk .
the column density of the slow wind is @xmath255 so it is completely shielded from the central radiation , both the x - rays and the uv photons .
acknowledgements : we would like to thank j.e
. drew for useful discussions .
the work presented in this paper was initiated while dp was a pparc research associate at imperial college , london and it was performed while dp held a national research council research associateship at nasa / gsfc .
computations were supported by nasa grant nra-97 - 12 - 055 - 154 .
= = 1 = .5 in hamann , f. , barlow , t.a . , cohen , r.d . ,
junkkarinen , v. , & burbidge , e.m .
1997 , in asp conf .
128 , mass ejection from active galactic nuclei , ed .
n. arav , i. shlosman , & r. weymann ( san francisco : asp ) , p. 19 | we present the results of axisymmetric time - dependent hydrodynamic calculations of line - driven winds from accretion disks in active galactic nuclei ( agn ) .
we assume the disk is flat , keplerian , geometrically thin , and optically thick , radiating according to the @xmath0-disk prescription .
the central engine of the agn is a source of both ionizing x - rays and wind - driving ultraviolet ( uv ) photons . to calculate the radiation force , we take into account radiation from the disk and the central engine .
the gas temperature and ionization state in the wind are calculated self - consistently from the photoionization and heating rate of the central engine .
we find that a disk accreting onto a @xmath1 black hole at the rate of @xmath2 can launch a wind at @xmath3 cm from the central engine .
the x - rays from the central object are significantly attenuated by the disk atmosphere so they can not prevent the local disk radiation from pushing matter away from the disk .
however in the supersonic portion of the flow high above the disk , the x - rays can overionize the gas and decrease the wind terminal velocity . for a reasonable x - ray opacity , e.g. , @xmath4 ,
the disk wind can be accelerated by the central uv radiation to velocities of up to 15000 @xmath5 at a distance of @xmath6 cm from the central engine .
the covering factor of the disk wind is @xmath7 .
the wind is unsteady and consists of an opaque , slow vertical flow near the disk that is bounded on the polar side by a high - velocity stream .
a typical column density through the fast stream is a few @xmath8 so the stream is optically thin to the uv radiation .
this low column density is precisely why gas can be accelerated to high velocities .
the fast stream contributes nearly 100% to the total wind mass loss rate of @xmath9 .
= = = 1=1=0pt = 2=2=0pt |
in this section , we summarize the derivation of eqs . through .
we consider an on - shell particle @xmath2 which decays to the particles @xmath1 and @xmath7 , both of which are on - shell ( see fig . [
fig : bldsystem ] ) .
we assume the masses of these particles are known and are @xmath9 , @xmath10 and @xmath11 , respectively .
we boost into the @xmath2 cm frame and calculate the cosine of the angle of @xmath1 with respect to the boost direction , which is given by @xmath22 where the subscript @xmath23 refers to the @xmath2 cm frame . for convenience , and without loss of generality ,
we take the @xmath24-direction to be in the same direction as @xmath25 , the momentum of @xmath2 . therefore , this can be rewritten as @xmath26 where @xmath27 refers to the z - component of the @xmath1 momentum .
[ [ section ] ] we will first consider the magnitude of eq . ,
which is given by @xmath28 for a massive @xmath1 we have latexmath:[\[\begin{aligned } @xmath12 indicates the component of @xmath30 that is transverse to @xmath31 .
we note that @xmath12 is invariant under a boost into the @xmath2 cm frame , and is therefore the same in the lab frame as in the @xmath2 cm frame . simplifying
, we have@xmath32 conservation of momentum in the @xmath2 cm frame leads to @xmath33 substituting this gives @xmath34 which completes the derivation of eq . .
we note that this is expressed in terms of lab - frame quantities and invariants only .
[ [ section-1 ] ] we now turn to the sign of @xmath35 .
we begin by boosting @xmath36 into the @xmath2 cm frame giving @xmath37 using the solution for @xmath38 from appendix [ sec : pdz ] and @xmath39 , we have @xmath40 where @xmath41 we have two sign possibilities given by the @xmath42 .
when this sign matches the sign of @xmath27 , we will call this the large " solution and use the symbol @xmath16 to denote it while if the sign is opposite that of @xmath27 , we will call this the small " solution and use the symbol @xmath17 to denote it .
if @xmath43 , since @xmath44 ( by definition of the @xmath24-direction ) , we must have the @xmath45 sign and be in the small solution . therefore according to eq .
@xmath46 where the @xmath47 refers to the fact that this is for @xmath43 . on the other hand , if @xmath48 , eq .
is the sign of the difference of two positive terms .
this sign is the same as the sign of the difference of these two positive terms squared giving @xmath49 using @xmath50 and @xmath51 , gives us @xmath52 by substituting eq . for @xmath53
this can be expanded and simplified to the form @xmath54 the large solution corresponds to the top sign choice , which is a minus sign , giving us @xmath55 which completes the derivation of eq . .
however , it still remains to calculate the small solution when @xmath48 .
taking the plus sign in eq . , dividing by the positive @xmath56 and squaring both ( positive ) terms we obtain @xmath57 plugging in @xmath58 , replacing @xmath59 with @xmath60 and then factoring @xmath60 out gives @xmath61 since this is the difference of two positive terms , we can take the positive square root of each term .
we can then multiply by the positive @xmath62 , substitute @xmath63 , simplify and divide by 2 to get @xmath64 with the function @xmath65 defined as @xmath66 we note from eq . that the maximum value of @xmath12 is given by @xmath67 from which it follows that @xmath68 however , in appendix [ sec : el > elcm ] we show that for the small solution when @xmath48 , we always have @xmath69 giving us @xmath70 putting eq . and
together gives @xmath71 however , we also show in appendix [ sec : el > elcm ] that @xmath72 .
plugging in eq
. finally gives us @xmath73 which completes the derivation of eq . .
we note that our final solutions depend on only the measured lab - frame energy of @xmath1 and the known masses .
we would next like to show that @xmath12 can be calculated directly from the masses of the particles and the energy of @xmath7 . for this section
, we note that eqs . through
do not depend on our choice of @xmath24-axis .
therefore , for this section , we choose a new reference frame for convenience and without loss of generality .
we take the @xmath24-axis to be along the direction of @xmath30 .
we further choose the plane of our interaction to be the @xmath74-@xmath24 plane . with this choice
, we have [ eq : allmomenta ] @xmath75 we can then use the unit vector that is normal to @xmath25 in the @xmath74-@xmath24 plane , @xmath76 to determine @xmath77 giving @xmath78 we can use the relation @xmath79 to obtain @xmath80 at which point , @xmath38 is the only unknown other than @xmath21 . to obtain this ,
we expand @xmath81 to get @xmath82 using eqs . and
, we can show that the denominator of eq .
is given by @xmath83 for the numerator of eq .
, we can use eq . to obtain @xmath84 putting this together ,
we finally obtain @xmath85 where @xmath86 is given by eq . .
this completes the derivation of eqs . and .
a special case of this kinematic variable was first used in @xcite where a new resonance in charged drell - yan ( d - y ) production of a charged lepton and a neutrino was considered .
it was there shown that once the mass of the new resonance is determined , although there is a two - fold ambiguity in the momentum of the neutrino which results in a two - fold ambiguity of the cm frame , it is nevertheless possible to reconstruct the full cm angular distribution of this new resonance . in this case , the spin and , in some cases , the parity violation of the couplings can be measured from this distribution . in this section
, we will summarize the salient points and discuss the effects of a finite width on these distributions . in figure
[ fig : dy ] , we show an illustrative diagram for the d - y resonance producing a charged lepton ( an electron or a muon ) and a neutrino .
we will call the resonance @xmath87 and assume its mass has been measured , for example , in a transverse mass distribution .
the mass of both the charged lepton and the neutrino are very small and will be taken as zero for clarity .
furthermore , due to momentum conservation , the total transverse momentum is zero with the result that the momentum of the charged lepton transverse to @xmath87 ( @xmath12 ) is equal to its momentum transverse to the beam direction ( @xmath88 ) , which is a measured quantity . as a result , we have the drell - yan process @xmath89 , where@xmath90 is some general charged resonance . ]
@xmath91 and @xmath92 in @xcite , it was shown that the large solution ( @xmath16 ) is sufficient to determine the spin of the resonance @xmath87 .
the parity - symmetric angular distribution can be reconstructed by dividing each @xmath93 bin by @xmath94 and taking the mirror image for the @xmath95 bins .
it was found that in the absence of cuts and for a very small width , the reconstructed angular distribution matched the true distribution perfectly .
it was also shown that the effect of cuts was to remove the large @xmath96 bins but did not affect the center of the distribution .
for example , it was shown that a cut of @xmath97gev only affected the @xmath98 bins .
the effect of a finite width was not explored in @xcite . in figure
[ fig : dylarge ] , we plot the reconstructed cm angular distribution using eq . with the effects of a finite width included .
cm angular distributions using the large solution ( eq . ) and reconstructing as outlined in the text .
the three curves are solid black for the true cm angular distribution , dashed blue for the reconstructed large solution when the width is 0.1% of the mass and dotted red when the width is 1% of the mass . from top
to bottom the three plots are for a scalar , vector and tensor @xmath87.,title="fig : " ] + cm angular distributions using the large solution ( eq . ) and reconstructing as outlined in the text .
the three curves are solid black for the true cm angular distribution , dashed blue for the reconstructed large solution when the width is 0.1% of the mass and dotted red when the width is 1% of the mass . from top to bottom the three plots are for a scalar , vector and tensor @xmath87.,title="fig : " ] + cm angular distributions using the large solution ( eq . ) and reconstructing as outlined in the text .
the three curves are solid black for the true cm angular distribution , dashed blue for the reconstructed large solution when the width is 0.1% of the mass and dotted red when the width is 1% of the mass . from top
to bottom the three plots are for a scalar , vector and tensor @xmath87.,title="fig : " ] the three plots are for a scalar ( top ) , vector ( middle ) and tensor ( bottom ) resonance , @xmath87 .
for each plot , we have included the true cm angular distribution in solid black , the reconstructed cm angular distribution for a 0.1% width in dashed blue , and the reconstructed cm angular distribution for a 1% width in dotted red .
we can see that for widths below 0.1% of the mass , there is very little alteration of the distribution . at 1% of the width , the modification of the distribution is still satisfactory .
only the bins for @xmath99 and @xmath100 are significantly affected .
however , for widths above 1% of the mass , the modification of the distribution starts becoming significant .
the small solution ( @xmath17 ) can be used to reconstruct the cm angular distribution as well , following the procedure outlined in @xcite .
although the reconstruction method is slightly more complicated , it has the advantage that the parity violation can also be reconstructed .
the effect of acceptance cuts was discussed in @xcite and is similar to that of the large solution .
we have analyzed the effect of finite widths and find it to be similar to the large solution case presented here .
we will now consider diagrams of the type shown in fig .
[ fig : antler ] . we will call this type of diagram an antler " diagram , following @xcite where the authors showed that the mass of @xmath2 and @xmath7 could , in principle , be measured in this process . an antler diagram for the pair production of two @xmath2 fields which then each decay to @xmath1 and @xmath7 .
the energy of each @xmath7 is known in this case as described in the text . ]
an analysis of this diagram shows that , given the masses of @xmath2 and @xmath7 , there is a two - fold ambiguity in the momenta of the two @xmath7s ( see app .
[ sec : twofold ambiguity ] ) . nevertheless , the energy of the two @xmath7s are known .
this can be seen by noting that @xmath101 where @xmath102 is the collision energy and @xmath103 determines which @xmath2 is being referred to
. conservation of energy , then , gives us @xmath104 since @xmath105 is a measured quantity , @xmath106 is known ( where @xmath103 determines which @xmath1 and @xmath7 is being considered ) . as we saw in sec .
[ sec : derivation ] , this is sufficient to apply our methods to this process , even though the @xmath107 cm frame can not be reconstructed . for definiteness , we will take @xmath1 to be a muon and allow @xmath2 and @xmath7 to take any consistent spin between @xmath108 and @xmath94 . the spins and charges of @xmath2 and @xmath7
are given in table [ table : particlenames ] .
.[table : particlenames]list of symbols used for the particles in our analysis with their charges and spins .
[ cols="^,^,^,^,^,^,^ " , ] we will further assume that @xmath7 is self - charge - conjugate .
we will take the two @xmath2s to be produced by their interaction with the photon and @xmath109 boson as in fig .
[ fig : ilcschannelmuon ] . pair production of @xmath2 followed by its decay to a muon and @xmath7 at the ilc . ] for this first illustrative use of our new kinematical variable at the ilc , we will take all new interactions to be parity symmetric .
the parity violating case will be considered in a later publication .
we used feynrules @xcite to implement these operators into calchep @xcite format .
we implemented our kinematic observables ( eqs . through ) using the usrmod functionality of calchep .
we then performed extensive simulations of all the consistent spin - combinations .
we note that the parity - conserving operators split into two categories .
the first category gives a symmetric cm angular distribution which only depends on the masses and the collision energy .
all dependence on the coupling constants drops out of the normalized differential cross - sections .
the majority of the parity - conserving operators fall into this category and were included in our analysis , in a model - independent way .
the resulting on - shell differential cross - sections were calculated analytically .
they have been included in table [ tab : analytic cm angular distributions ] and will be discussed further below . a complete list of the operators in this category can be found in app .
[ sec : effective operators ] .
a discussion of the derivation of the analytic differential cross - sections coming from these operators is included in appendix [ sec : cm analytic derivation ] .
the second category of parity - conserving operators also gives a symmetric cm angular distribution , however the expressions for the differential cross - sections are more complicated and depend on the values of the coupling constants .
in particular , the values of @xmath110 , @xmath111 , @xmath112 and @xmath113 in table [ tab : analytic cm angular distributions ] depend non - trivially on the coupling constants in the second category of parity - conserving operators ( a discussion of the operators from this category is also included in appendix [ sec : effective operators ] ) .
since our main purpose in the present article is to show that our kinematical variables faithfully reproduce the true cm angular distribution , for simplicity , we focus on the parity - conserving operators from the first category . in order to ascertain the power of our kinematic variables , we would like to compare its results with the true " cm angular distribution . since the two @xmath7s in this process are identical , we numerically calculated the true cm angular distribution in two ways . first , in the process
@xmath114 , we used the monte carlo information to determine which @xmath7 to pair with @xmath115 .
we then boosted into this frame to calculate the true cm angular distribution .
second , we simulated the process @xmath116 , where the final @xmath117 was not decayed .
we then boosted into the @xmath118 rest frame and determined the cm angular distribution .
in all cases , we found agreement between these methods , as expected .
we simulated the _ true
_ cm angular distribution for every consistent spin - combination . in all narrow width cases in the absence of cuts
, we found perfect agreement between the analytic expressions from table [ tab : analytic cm angular distributions ] and the simulations of the true cm angular distributions .
this agreement can be seen in fig .
[ fig : classes plots ] where the analytic formulas are used to produce the solid green curves and the monte carlo simulation of the _ true _ solution is plotted in solid black .
the illustrative values @xmath119gev , @xmath120gev and @xmath121gev were used to create this figure .
as expected , all the angular distributions are symmetric since we only considered parity - symmetric operators in this analysis .
llll & @xmath122 & @xmath123 & + i & @xmath108 & & + i & @xmath124 & & + ii & @xmath125 & @xmath124 & + + & & & @xmath126 + & & & @xmath127 + & & & @xmath128 + iiia & @xmath125 & @xmath129 & + + & & & @xmath130 + & & & @xmath131 + & & & @xmath132 + iiib & @xmath129 & @xmath108 & + + & & & @xmath133 + & & & @xmath134 + & & & @xmath135 + iiic & @xmath129 & @xmath125 & + + & & & @xmath136 + & & & @xmath137 + & & & @xmath138 + iiid & @xmath129 & @xmath94 & + + & & & @xmath139 + & & & @xmath140 + & & & @xmath141 + & & & @xmath142 + iva & @xmath94 & @xmath124 & + + & & & @xmath143 + & & & @xmath144 + & & & @xmath145 + & & & @xmath146 + ivb & @xmath94 & @xmath129 & + + & & & @xmath147 + & & & @xmath148 + & & & @xmath149 + & & & @xmath150 + + ( i ) + ( ii ) + ( iiia ) + ( iiib ) + + ( iiic ) + ( iiid ) + ( iva ) + ( ivb ) we also simulated the full signal final state @xmath151 and calculated our kinematical variable using eqs . through using _ only _ the momentum of the muon .
the large solution ( eq . [ eq : largesolution ] ) is shown as the dashed blue line in fig .
[ fig : classes plots ] . as expected , it is nonzero only for @xmath152 and exactly double the true distribution .
we can reconstruct the true solution by dividing this distribution by @xmath94 and taking the mirror image on the @xmath153 side .
this has been done and is plotted as the red dotted curves in fig . [
fig : classes plots ] . as can be seen
, the red dotted curves agree perfectly with the true solutions in solid green and solid black .
the small solution would give a similar result , but would require a slightly more complicated reconstruction method .
however , because the distribution is fundamentally symmetric in the present article , the sign is not helpful .
therefore , since the large solution is sufficient to get exact agreement with the true solution , we will not discuss the small solution further in this section .
we have found that the various spin - combinations fall into four major classes : i ) flat distributions , ii ) parabolic distributions with negative concavity , iii ) parabolic distributions with positive concavity , and iv ) m " shaped distributions .
this classification corresponds with the first column of table [ tab : analytic cm angular distributions ] and the labels in figs .
[ fig : classes plots ] through [ fig : isr plots ] and described in detail below . the first class is given in row ( i ) of table [ tab : analytic cm angular distributions ] and shown in fig . [ fig : classes plots]i .
it is a flat line with no @xmath154 dependence .
this class not only includes the case where @xmath2 has spin @xmath108 , but also includes the cases where @xmath2 has spin @xmath124 .
the reason for this is that the inherent parity symmetry of these angular distribution requires equal contributions from @xmath155 and @xmath156 for any given @xmath4 , @xmath5 and @xmath6 . in the case of a spin-@xmath124 @xmath2 , the wigner @xmath157-functions are @xmath158 and @xmath159 . as a result , since the differential cross section is proportional to the linear combination of the squares of the wigner @xmath157-functions ( see app .
[ sec : wigner d - functions ] ) , the normalized differential cross section in this case is given by @xmath160 = \frac{1}{2}\ .\ ] ] the second class of cm angular distributions is given by a concave negative parabola as seen in fig .
[ fig : classes plots]ii .
there is only one spin combination that gives this distribution and that is a spin-@xmath125 @xmath2 decaying to a spin-@xmath124 @xmath7 .
the wigner d - functions for spin-@xmath125 are @xmath161 , @xmath162 and @xmath163 , whose squares are quadratic in @xmath154 , resulting in a parabola .
the expression for the normalized differential cross section in this case is given in row ( ii ) of table [ tab : analytic cm angular distributions ] .
we see that the concavity is proportional to @xmath164 and that , while always negative in this case , is increased in size as the mass difference between @xmath2 and @xmath7 is increased and also as the difference between the collision energy @xmath102 and the mass of @xmath2 is increased .
although we do not have any control over the mass difference between @xmath2 and @xmath7 , we do have some control over the collision energy .
if a distribution appears to be nearly flat , the collision energy should be increased to determine whether this property changes .
this dependence of the shape on the collision energy is a feature of all the classes discussed here except , of course , class i , the flat distribution .
the third class of distributions is given by concave up parabolas as shown in fig .
[ fig : classes plots]iii .
there are four spin - combinations that give this class of distributions .
the first is a spin-@xmath125 @xmath2 decaying to a spin-@xmath129 @xmath7 ( see fig . [
fig : classes plots]iiia ) .
the second , third and fourth are , respectively , a spin-@xmath129 @xmath2 decaying to a spin-@xmath108 @xmath7 ( see fig . [ fig : classes plots]iiib ) , a spin-@xmath125 @xmath7 ( see fig . [
fig : classes plots]iiic ) and a spin-@xmath94 @xmath7 ( see fig . [
fig : classes plots]iiid ) .
initially , when @xmath2 is spin-@xmath129 , the wigner d - functions squared are cubic in @xmath154 .
however , since these distributions are inherently parity symmetric , the odd terms cancel ( as in the previously discussed spin-@xmath124 case ) and we are left with expressions quadratic in @xmath154 .
we further note that , although the fig .
[ fig : classes plots]iiia distribution appears to be equal to the fig . [ fig : classes plots]iiib distribution and the fig . [ fig : classes plots]iiic distribution appears to be equal to the fig . [
fig : classes plots]iiid distribution , this is only a coincidence of the illustrative masses and collision energies we used . if we , instead , used @xmath165tev , @xmath120gev and @xmath166gev , the four distributions would all be separated as in fig . [
fig : iii splitting]iii .
class iii and iv cm angular distributions for the values @xmath165tev , @xmath120gev and @xmath166gev . in the case of iii , the solid black line is for class iiia , the blue dashed line is for class iiib , the red dot - dashed line is for class iiic and the orange dotted line is for class iiid . in the case of iv ,
the solid black line is for case iva while the dashed blue line is for case ivb.,title="fig : " ] + ( iii ) + class iii and iv cm angular distributions for the values @xmath165tev , @xmath120gev and @xmath166gev . in the case of iii
, the solid black line is for class iiia , the blue dashed line is for class iiib , the red dot - dashed line is for class iiic and the orange dotted line is for class iiid . in the case of iv ,
the solid black line is for case iva while the dashed blue line is for case ivb.,title="fig : " ] + ( iv ) as we see in table [ tab : analytic cm angular distributions ] , in each of these cases , the strength of the concavity is dependent on both the mass splitting between @xmath2 and @xmath7 as well as on the collision energy .
we can not choose the mass splitting .
however , whatever the mass splitting , we can , in principle , choose higher collision energies to accentuate the shape of each of these cases .
the final class of distributions is given by an
m " shape which is quartic in @xmath154 .
this comes from a spin-@xmath94 @xmath2 decaying to either a spin-@xmath124 @xmath7 ( see fig . [
fig : classes plots]iva ) or a spin-@xmath129 @xmath7 ( see fig . [
fig : classes plots]ivb ) . as in the previous classes , the quadratic and quartic terms in the angular dependence ( see rows iva and
ivb in table [ tab : analytic cm angular distributions ] ) are proportional to @xmath167 , therefore , the shape can be accentuated by increasing the collision energy .
we give an example of this in fig .
[ fig : iii splitting]iv , where @xmath165tev , @xmath120gev and @xmath166gev . in the previous subsection
, we took the width of @xmath2 to be very small ( @xmath168% of the mass ) and found perfect agreement between the _ true _ cm angular distributions and our reconstructed distributions .
however , the derivation of our kinematical variables , eqs . through , relied heavily on @xmath2 being on - shell , and therefore having a narrow width .
furthermore , off - shell , even the true cm angular distribution could be modified by off - shell polarization vectors .
when the spin of b is greater than 1 , the off - shell polarization effect is model dependent . in this subsection , we would like to estimate the total effect on our reconstruction technique of having a more sizable width . in fig .
[ fig : fw plots ] , we have plotted our reconstructed cm angular distribution when the width of @xmath2 is @xmath125% in solid black and @xmath169% in dashed blue . for comparison , we have also plotted the _
true _ on - shell cm angular distribution in solid green . for spins greater than 1 ,
we have used the standard propagators that are built into calchep @xcite . + ( i )
+ ( ii ) + ( iiia ) + ( iiib ) + + ( iiic ) + ( iiid ) + ( iva ) + ( ivb ) in general , we find that the effect of the width is almost negligible when @xmath125% of the mass and only begins to be sizable for some spin combinations when approximately @xmath169% of the mass . however , even at @xmath169% , the agreement with the theoretical curve is still quite good in most cases .
the main effect is to reduce the cm angular distribution for large @xmath170 and , to a lesser extent , to increase the distribution for small @xmath170 . in the case of the theoretically concave negative distribution ( see fig . [
fig : fw plots]ii ) , the effect of both a @xmath125% and @xmath169% width is negligible . in the case of the theoretically flat distribution ( see fig . [
fig : fw plots]i ) , the effect of a @xmath125% width is to slightly reduce the edges when @xmath171 while the enhancement in the center is not very great .
when the width becomes @xmath169% , the effect begins to be significant but depends on which spin combination is being considered .
the distributions for spin-@xmath108 @xmath2 and for spin-@xmath124 @xmath2 decaying to a spin-@xmath108 or spin-@xmath125 @xmath7 , the effect is still small with the edges only being strongly affected for @xmath172 and the center not being strongly enhanced .
however , for a spin-@xmath124 @xmath2 decaying to a spin-@xmath94 @xmath7 , the effect is much larger with the shape changing significantly to look nearly like a parabola with negative concavity . however , since the negative concavity parabola ( spin-@xmath125 @xmath2 decaying to a spin-@xmath124 @xmath7 ) is not significantly affected by a finite width , it should not be difficult to distinguish these cases . in the case of the concave positive distributions ( see fig . [
fig : fw plots]iii ) , the effect is small and mainly at the edges . for a @xmath125% width ,
the reduction is visible for @xmath171 while the central enhancement is minimal . for a @xmath169% width ,
the reduction is visible for @xmath172 .
again , the enhancement is minimal . for the case of a spin-@xmath94 @xmath2 decaying to a spin-@xmath124 @xmath7 ( see fig . [
fig : fw plots]iva ) , the effect of a @xmath125% width is minimal , however , a @xmath169% width does change the distribution .
although it retains the m " shape , the outer edges are approximately @xmath173% lower while the center is approximately @xmath173% higher . in the case of a spin-@xmath94 @xmath2 decaying to a spin-@xmath129 @xmath7 ( see fig . [
fig : fw plots]ivb ) , the effect of even a @xmath169% width is very small with a slight decrease towards the edges .
we also considered the effects of initial state radiation ( isr ) , beamstrahlung and basic cuts in an electron - positron collider environment on our distributions .
the cuts we used were @xmath174 to ensure detection , where @xmath175 and @xmath176 are , respectively , the polar angle and the energy of the electron in the lab frame , and @xmath177 to remove the photon t - channel induced background from @xmath178 where the final electron - positron pair are missed @xcite , where @xmath179 is the missing transverse momentum .
we display these effects in fig .
[ fig : isr plots ] .
+ ( i ) + ( ii ) + ( iiia ) + ( iiib ) + + ( iiic ) + ( iiid ) + ( iva ) + ( ivb ) the theoretical curves in the absence of isr , beamstrahlung and cuts is shown in solid green .
the effect of including isr and beamstrahlung is shown as the solid black lines and the additional effect of our basic cuts is shown by the dashed blue lines .
the solid black and dashed blue lines are reconstructed from the large solution , eq . .
in all cases , the solid black lines and dashed blue lines coincide showing that these basic cuts do not have a significant impact on the cm angular distributions .
of course , further cuts will be required to separate the signal from the background , but we leave that to a separate study .
isr and beamstrahlung , on the other hand , do modify these distributions .
the modification is similar to that of the finite widths , but more extreme showing that the effect of finite widths is probably subdominant for measuring the cm angular distribution in this process .
the smearing due to isr and beamstrahlung appears to be the dominant effect modifying the cm angular distributions from their theoretical form .
it reduces the edges and enhances the center of these distributions . for the flat distributions ( fig .
[ fig : isr plots]i ) , the edges are significantly reduced by approximately 60% while the center is only enhanced by approximately 10% .
in contrast to the effect of a finite width , the new distribution does not look parabolic and should be easily distinguishable . for the concave negative parabola distribution ( fig .
[ fig : isr plots]ii ) , the effect is small with a significant deviation from the theoretical curve only for @xmath180 . for the concave positive parabola distributions ( fig .
[ fig : isr plots]iii ) , the effect is to alter the distribution to look more like the m " distributions .
this could make distinguishing these two cases more difficult , however , it can be seen that the position of the peaks in the two cases is different as is the shape of the shoulder and the concavity of the central region . with sufficient collision energy and luminosity , these cases should also be distinguishable . for the theoretically m " shaped distribution ( fig .
[ fig : isr plots]iv ) , the effect is to decrease the edges further while not changing much the central region . in this section ,
we have illustrated the use of our kinematical variables for reconstructing the cm angular distribution in the process @xmath181 .
we have found excellent agreement between our reconstruction and the true cm angular distribution , even in the presence of a small finite width .
additionally , we showed that the curvature of the distributions depended on the collision energy , which is in principle controllable at the ilc .
we have further found that although isr and beamstrahlung modify the distributions , this effect is predictable with monte carlo .
it appears as a reduction at the edges and a small enhancement in the center of the distribution . for most of the classes ,
the essential shape can still be seen . therefore , in principle , it should be possible with sufficient luminosity to distinguish all the spin combinations using our kinematical variable except for those within case i where @xmath2 is spin-0 or -@xmath124 or between spin combinations within the same class when the masses and collision energy conspire to make the distributions identical . to end this section , we would like to discuss a few areas where the analysis of these kinematic variables applied to antler processes can be extended .
in this work , we have only considered parity symmetric s - channel production of the two @xmath2s . in certain cases , parity violating operators may be important . in these cases , we may be able to further separate the spin combination cases as well as get information about the parity violating nature of the couplings by using the small solution .
furthermore , in some models , the t - channel may be important .
this is certainly true in the sm , where the process @xmath182 has an important contribution from a neutrino in the t - channel .
we have found that , in this sm process , the cm angular distribution when the t - channel diagram is included is sensitive to chiral couplings and that parity violation can be seen in the small solution . a full analysis of the effect of parity violating operators and of the t - channel production will be carried out in a future publication .
we have assumed in this section that @xmath7 was self - charge - conjugate .
however , this does not have to be the case .
@xmath7 does not have to be its own antiparticle .
although we have not done a full analysis of this possibility , we have found that the cm angular distribution in the case of the sm process @xmath182 is the same whether the neutrinos are dirac or majorana . for our illustration ,
we used muons in the final state , however , our method would work just as well with jets .
the benefit of using muons is that they can be distinguished by their sign and their momentum can be measured more precisely than jets .
however , in some models , the branching ratio to jets could be much higher .
although the jets are indistinguishable , it does not matter for the symmetric distributions we have presented here because each jet generates the same distribution . in this case , the cm angular distribution coming from the large solution ( eq . ) should be separately binned for each jet .
the final distribution will have the same shape as that for an individual jet but will be twice as high in each bin . in this work ,
we have only considered the intermediate particle @xmath2 decaying to a two - body final state @xmath1 and @xmath7 , however , it could also decay to multiple visible particles and @xmath7 . if the multiple visible particles come from another intermediate state that decays ( such as a w decaying to jets ) then there is no change to the results presented here
. the mass and spin component of @xmath1 are just those of the extra intermediate state ( such as the w ) . however ,
if , on the other hand , there is no intermediate state decaying to the multiple visible particles , @xmath183 can still be reconstructed for each event , but the cm angular distribution depends on the invariant mass @xmath10 which changes event to event .
it should , nevertheless , be possible to generate a series of cm angular distributions for each invariant mass range of the visible particles . however , the situation is further complicated by the fact that the component of the angular momentum of the visible particles will not simply be the component of the spin of a single particle anymore , further complicating the interpretation of the cm angular distributions .
in this work , we have constructed a new kinematical variable that gives , under certain circumstances , the cm angular distribution , @xmath184 , of a decaying particle @xmath2 when one of its daughter particles @xmath7 is not detected . among the requirements for this method
are that the masses of the mother particle @xmath2 as well as both daughter particles @xmath1 and @xmath7 must be known .
additionally , we have shown that the magnitude of @xmath184 only depends on the component of the momentum of the measured particle @xmath1 which is transverse to the momentum of @xmath2 , which we call @xmath12 , as can be seen in eq . .
we have also shown that @xmath12 is uniquely related to the energy of the missed particle @xmath7 , @xmath21 , as seen in eqs . and .
therefore , if either @xmath12 or @xmath21 can be determined event by event , then the magnitude of @xmath184 is known , event by event , unambiguously .
the sign of our observable , on the other hand , has an inherent two - fold ambiguity .
we find that it can be consistently split into two solutions which we call the large and small solutions , based on the size of the component of @xmath185 which is parallel to @xmath30 .
we find that the large solution always has a negative sign , as in eq . while the small solution s sign depends on the energy of the measured particle @xmath1 , as in eq . .
although this sign can not be determined event by event , the distributions of the large and small cm angular distributions , eq . and eq . , respectively ,
contain the information required to reconstruct the true @xmath184 in many situations .
the large solution gives the negative absolute value of the distribution of @xmath184 , from which it is easy to reconstruct the full symmetrized version of the @xmath184 distribution by dividing it in half and taking the mirror image on the @xmath186 side .
although this distribution can not tell us anything about the parity violation in the true @xmath184 distribution , it can give us the full symmetrized dependence of the differential cross - section on @xmath184 and therefore the spin - combinations of the mother and daughter particles which can produce that distribution .
in many cases , this is already sufficient to determine the spin of @xmath2 and @xmath7 , where the spin of @xmath1 is already known .
although the full structure of our kinematical variable was first described in the present work , a special case of this kinematical variable was applied to d - y production of a charged lepton and a neutrino where the spin of the intermediate resonance was shown to be unambiguously determined by the large solution @xcite . on the other hand
, the small solution contains nontrivial information about the sign of @xmath184 .
although it does not agree with the true sign event by event , its distribution contains the clear signatures of the parity violation present in the true distribution . in @xcite
, it was shown that in the special case of d - y production of a charged lepton and a neutrino , a simple reconstruction technique could be applied to the small solution to fully reconstruct the true cm angular distribution , including its parity violating features , almost exactly .
it was , further , shown that this reconstruction technique was universal and did not depend on the spin of @xmath2 or the parity - violation .
in other words , the reconstruction technique could be applied blindly and give the correct results in all spin cases . moreover
, it was also shown that the parity violation could be determined directly from the small solution without applying the reconstruction technique . in @xcite , it was also shown that acceptance cuts , and even the rather large @xmath187gev cut only affected the @xmath188 edges of the distributions , but left the majority of the distribution unaffected , preserving its power . in the present work , we extended this to consider the effect of a finite width .
we found that an approximately @xmath125% width and smaller widths only affected the @xmath189 edges and the @xmath190 central region , but left the rest of the distribution unchanged , as seen in fig .
[ fig : dylarge ] .
this shows that this variable works quite well for realistic d - y charged resonance production at the lhc . in the present work , we applied our kinematical variables , for the first time , to the antler process @xmath181 at the ilc where @xmath7 was taken to be a self - charge - conjugate dark matter particle .
we showed that @xmath21 is known in this case , therefore @xmath183 is known , event by event , unambiguously , as described above .
we focused , in this article , on only parity - symmetric operators , therefore all our angular distributions were inherently symmetric .
we calculated analytically the dependence of the differential cross - section on @xmath184 and included it in table [ tab : analytic cm angular distributions ] .
we found agreement of the true cm angular distribution with these formulas in all cases .
the cm angular distributions split up into four classes . the first class ( denoted by i in table [ tab : analytic cm angular distributions ] and figs .
[ fig : classes plots ] through [ fig : isr plots ] ) is given by a flat distribution and , unfortunately , includes both the cases where @xmath2 is spin-@xmath108 and spin-@xmath124 . the second class ( denoted by ii in table [ tab : analytic cm angular distributions ] and figs .
[ fig : classes plots ] through [ fig : isr plots ] ) is given by a concave negative parabola and includes only the case where @xmath2 is spin-@xmath125 and @xmath7 is spin-@xmath124 . the third class ( denoted by iii in table [ tab : analytic cm angular distributions ] and figs .
[ fig : classes plots ] through [ fig : isr plots ] ) is given by a concave positive parabola and includes the case where @xmath2 is spin-@xmath125 and @xmath7 is spin-@xmath129 and the cases where @xmath2 is spin-@xmath129 and @xmath7 is spin-@xmath108 , @xmath125 or @xmath94 .
although their distributions are all concave positive parabolas , we show that with appropriate masses and collision energy , they can be separated and distinguished as in fig .
[ fig : iii splitting]iii . the last class ( denoted by iv in table [ tab : analytic cm angular distributions ] and figs .
[ fig : classes plots ] through [ fig : isr plots ] ) is given by a m " shape and includes the cases where @xmath2 is spin-@xmath94 and @xmath7 is spin-@xmath124 or @xmath129 .
again , we show that for appropriate parameters , these can be distinguished as in fig .
[ fig : iii splitting]iv
. we then showed that the large cm angular distribution after a simple reconstruction , gives exact agreement with the true distribution in the narrow width limit and in the absence of isr and beamstrahlung , as shown in fig .
[ fig : classes plots ] .
if the width is not infinitesimal , on the other hand , we found a small reduction on the edges and , to a lesser extent , a small enhancement in the center of the distribution occured .
however , for a @xmath125% width , we found the effect to be practically negligible while a @xmath169% width made a noticeable effect but left the shape of the distributions largely intact for most spin combinations , as seen in fig .
[ fig : fw plots ] .
the effect of isr and beamstrahlung , on the other hand , was much more pronounced .
it also appeared as a reduction on the edges and , to a lesser extent , an enhancement in the center .
however , as can be seen in fig .
[ fig : isr plots ] , the shape of the true distribution is still clearly visible for most of the spin combinations .
furthermore , the effect of isr and beamstrahlung can be well modeled . from these results
, we see that our method should work quite well in this process and that as long as @xmath2 is spin-@xmath125 or higher and the masses do nt conspire to make the distributions within a class identical , it should be possible to determine the spin of both @xmath2 and the dark matter particle @xmath7 using our kinematical variables in this process at the ilc .
_ acknowledgements _ n.d.c . was supported in part by pitt pacc and the u.s .
department of energy under grant no .
de - fg02 - 95er40896 .
d.s . was supported in part by pitt pacc , the pennsylvania space grant consortium research scholarship and the dietrich school of arts and sciences summer undergraduate research award for independent research .
we would like to thank tao han , adam leibovich and ayres freitas for their encouragement , helpful discussions and support during the completion of this project .
we would also like to thank the university of granada high energy theory group for their hospitality during our visit where part of this research was completed .
diagram of a scattering process dominated by an on - shell @xmath2 which decays to @xmath1 and @xmath7 . ] in this appendix , we give a very rough description of the dependence of the differential cross section on the wigner d - functions .
consider a scattering process that is dominated by diagrams where an intermediate particle @xmath2 is on - shell and decays to @xmath1 and @xmath7 as in fig .
[ fig : wigner diagram ] .
since @xmath2 is on - shell , its propagator numerator is equal to a sum over the spin - z components of dyads of its polarization vector , as in @xmath191 where @xmath192 is @xmath2 s propagator numerator and @xmath193 is @xmath2 s polarization vector for spin - z component @xmath194 .
( we use @xmath195 here to represent the polarization vector or spinor for any spin . ) using this , the amplitude can be written as @xmath196 where @xmath197 contains everything not explicitly written in this equation , including the polarization vectors of all the other external states , the polarization vector for @xmath2 and the propagator denominator for @xmath2 .
@xmath198 represents the vertex factor for the @xmath199 system and may connect to the polarization vectors and also may connect to @xmath197 .
we will consider the @xmath199 system in the @xmath2 cm frame .
we take the @xmath24-direction to be along @xmath25 and take @xmath194 to be the component of @xmath2 s spin along this direction .
since this system consists solely of @xmath2 before the decay , the total angular momentum is equal to the spin of @xmath2 .
because total angular momentum is conserved , the final state @xmath200 must have the same total angular momentum .
if we measure the final total angular momentum component along @xmath30 , it will be equal to the difference of the helicities of @xmath1 and @xmath7 , which we will call @xmath201 and @xmath202 , respectively .
although orbital angular momentum in the @xmath200 system can contribute to the total angular momentum , its component is perpendicular to @xmath30 and , therefore , will not contribute to the component of the total angular momentum along that direction .
since total angular momentum is conserved , the dependence on this angle is simply the quantum mechanical overlap between these bases , @xmath203 where @xmath204 is the component of the angular momentum operator perpendicular to the @xmath199 system .
these functions are called the wigner d - functions . with this , in the @xmath2 cm frame , we have @xmath205 putting these things together gives @xmath206 where @xmath207 absorbs all other factors . after squaring , averaging over initial spins , summing over final spins , multiplying by phase factors and integrating over all other momentum factors , we have @xmath208 we see that the dependence on this angle is restricted by conservation of angular momentum to be proportional to a sum of squares of the wigner @xmath209-functions , where @xmath210 , @xmath211 or @xmath212 and @xmath213 , independent of the model .
the coefficients @xmath214 , on the other hand , are dependent on the couplings of the model as well as the polarization of the beams .
in this appendix , we derive @xmath38 , the z - component of the @xmath7 momentum . since the @xmath24-axis was defined to be in the same direction as the momentum of @xmath2 , the net momentum transverse to this direction must be zero . without loss of generality
, we will take @xmath12 to be along the @xmath74-axis .
we can then define the momentum of @xmath1 and @xmath7 as @xmath215 the invariant mass of @xmath2 is then given by @xmath216 this equation can be solved for @xmath38 giving @xmath217 where @xmath218 recalling the relation for @xmath62 ( eq . ) , we have @xmath219 we note that our result does not depend on the choice of @xmath74- and @xmath220-axis .
in this appendix , we assume the small solution . using eq .
when @xmath48 , we have @xmath221 for the small solution . plugging in @xmath56 ( see eq . ) , using @xmath51 and factoring out the positive @xmath222 gives @xmath223 on the other hand , when @xmath43 , we have @xmath224 for the small solution , which can be rewritten as @xmath225 following the same manipulations as above brings this to the form @xmath226 or @xmath227 from these two results , we learn that @xmath228
in this section , we consider processes at the ilc of the type shown in fig .
[ fig : antler ] , dubbed antler diagrams . following @xcite
, we will assume that the masses of @xmath2 and @xmath7 are known and show that there is only a discrete ambiguity in the momenta of the two @xmath7s .
we begin by noting that there are eight unknowns in the momenta of the two @xmath7s and count the constraints . due to conservation of energy and the symmetry of the diagram ,
we find : @xmath229 where @xmath103 refers to which @xmath2 and @xmath230 in fig .
[ fig : antler ] .
this implies the two linear equations @xmath231 and completely specifies the energy of the two @xmath7s .
we also have the conservation of three - momenta which gives the three linear equations @xmath232 for the mass of b , we have @xmath233 which reduces to @xmath234 after plugging in eq . , this gives us the two following linear equations @xmath235 finally , we have the mass of d which , after substituting eq .
, gives the two quadratic equations @xmath236 all together this makes nine equations for eight unknowns .
seven are linear and two are quadratic , one of which is linearly dependent .
this leaves us with two discrete solutions . for further details ,
see @xcite .
in this appendix we provide a list of all the parity - conserving operators used in our analysis , except those already described in the sm . the naming conventions for the fields
are given in table [ table : particlenames ] .
as described in sec . [ sec : antlers ] , we included the majority of parity - conserving operators in our analysis in a model independent way .
there were only a few parity - conserving operators that we did not include and we specify them as they come up in this appendix .
these operators will be considered in a future publication .
these operators follow directly from qed , where we replace the partial derivative in the kinetic term with the covariant derivative , denoted here by @xmath238 , to ensure that our lagrangians are manifestly qed invariant . for spin-0
, we have @xmath239 for spin-@xmath124 , we have @xmath240 for spin-@xmath125 , we have @xmath241 the lagrangian for a spin-3/2 field was first reported by rarita and schwinger @xcite . however , it is now customary to write the lagrangian in a way that is more compact , although entirely equivalent . in our work ,
we use the lagrangian given in @xcite , @xmath242 for spin-@xmath94 , we have @xcite @xmath243 where @xmath244 we could have also included a term of the form @xmath245 which is also parity conserving , however we find that this operator gives a more complicated form of the symmetric cm angular distribution .
specifically , @xmath246 , @xmath247 , @xmath248 and @xmath249 from table [ tab : analytic cm angular distributions ] depend on the value of @xmath250 if non - zero . for simplicity , in the current article
, we choose to set @xmath251 so that @xmath246 , @xmath247 , @xmath248 and @xmath249 depend only on the masses and collision energy . for convenience
, we implemented the photon interactions of the previous subsection with the hypercharge gauge boson .
thus , the covariant derivative is of the form @xmath253 therefore , the operators of the previous subsection induce interactions with the z boson as well .
in addition to these operators , we also included the lagrangian terms as described below . for spin-@xmath108
, we include the lagrangian @xmath254 for spin-@xmath124 , we have @xmath255 for spin-@xmath125 , we include @xmath256 we could have also included the operator @xmath257 which is also parity conserving , however we find that , as for the case where @xmath2 is spin-2 coupling to a photon , this operator gives a more complicated form of the symmetric cm angular distribution .
specifically , @xmath258 , @xmath259 , @xmath260 , @xmath261 , @xmath262 and @xmath263 from table [ tab : analytic cm angular distributions ] depend on the value of @xmath264 if non - zero . for simplicity , in the current article , we choose to set @xmath265 so that @xmath258 , @xmath259 , @xmath260 , @xmath261 , @xmath262 and @xmath263 depend only on the masses and collision energy . for spin-@xmath129
, we include the operator @xmath266 for spin-@xmath94 , we include @xmath267 we could have also included the operators @xmath268 which are also parity conserving , however we find that , as for the case where @xmath2 is spin-1 , these operators give a more complicated form of the symmetric cm angular distribution .
specifically , @xmath246 , @xmath247 , @xmath248 and @xmath249 from table [ tab : analytic cm angular distributions ] depend on the values of @xmath269 and @xmath270 if non - zero . for simplicity , in the current article , we choose to set @xmath271 so that @xmath246 , @xmath247 , @xmath248 and @xmath249 depend only on the masses and collision energy .
we begin with the lagrangian for the case where @xmath2 is spin-@xmath108 , @xmath274 for the case in which @xmath2 is spin-@xmath124 , the lagrangian reads @xmath275 in the case where @xmath2 is spin-@xmath125 , we have @xmath276 for a spin-@xmath129 @xmath2 , we have @xmath277 finally , we have for the spin-@xmath94 particle @xmath278
in this section , we describe the derivation of the formulas in table [ tab : analytic cm angular distributions ] . using the calchep implementation of our new particles @xmath2 and
@xmath7 along with their interactions , we generated the analytic expressions for the squared amplitude of the diagram shown in fig .
[ fig:2->3 antler ] . from there , we exported these expressions to mathematica .
( in the calchep symbolic interface , after squaring the diagrams choose symbolic calculations " followed by mathematica code " .
the resulting mathematica files will be in the results directory @xcite . ) at this point , we had the squared amplitude in terms of masses , collision energy and inner products of the external momenta .
we note that we included the photon - photon squared diagram , the z - z squared diagram and the interference photon - z and z - photon squared diagrams in our calculation .
we removed the z width from the calculation since its momentum was far off - shell , however we included the width of the @xmath2 in the calculation , which was necessary since we assume @xmath2 is on - shell for these formulas . by conservation of momenta
, @xmath280 can be removed from these expressions so we only need to deal with @xmath281 through @xmath282 .
since each of the inner products is an invariant , we can do this calculation in any reference frame .
we choose the @xmath24-direction to be along the direction of @xmath283 so that in the lab frame , @xmath284 .
we then boost into the @xmath2 cm frame . in this frame , @xmath285 where , by the usual conservation of momentum arguments , @xmath286 the initial momenta are known in the lab frame .
they are @xmath287 where the angle @xmath288 is with respect to the direction of @xmath25 . after boosting into the @xmath2 cm frame , we have @xmath289 where @xmath290 and @xmath291 .
we plugged these expressions for the momenta into the squared amplitude formulas . since the differential cross section is equal to this squared amplitude times factors that are independent of these angles , we next integrated our expression over the angles @xmath292 and @xmath293 .
this left us with an expression in terms of only @xmath294 and @xmath0 .
we finally normalized this expression by dividing by the same expression integrated over @xmath0 .
this gave us the normalized differential cross - sections found in table [ tab : analytic cm angular distributions ] .
a. bosma , astron .
j. * 86 * , 1825 ( 1981 ) .
a. refregier , ann .
astrophys . * 41 * , 645 ( 2003 ) [ astro - ph/0307212 ] .
j. a. tyson , g. p. kochanski and i. p. dellantonio , astrophys .
j. * 498 * , l107 ( 1998 ) [ astro - ph/9801193 ] .
k. a. olive , g. steigman and t. p. walker , phys .
rept . * 333 * , 389 ( 2000 ) [ astro - ph/9905320 ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl . * 192 * , 18 ( 2011 ) [ arxiv:1001.4538 [ astro-ph.co ] ] .
d. clowe , m. bradac , a. h. gonzalez , m. markevitch , s. w. randall , c. jones and d. zaritsky , astrophys . j. * 648 * , l109 ( 2006 ) [ astro - ph/0608407 ] .
h. goldberg , phys . rev . lett .
* 50 * , 1419 ( 1983 ) [ erratum - ibid .
* 103 * , 099905 ( 2009 ) ] .
j. r. ellis , j. s. hagelin , d. v. nanopoulos , k. a. olive and m. srednicki , nucl .
b * 238 * , 453 ( 1984 ) .
h. -c .
cheng , j. l. feng and k. t. matchev , phys .
lett . * 89 * , 211301 ( 2002 ) [ hep - ph/0207125 ] .
g. servant and t. m. p. tait , nucl .
b * 650 * , 391 ( 2003 ) [ hep - ph/0206071 ] . c. -w .
chiang , n. d. christensen , g. -j .
ding and t. han , phys .
d * 85 * , 015023 ( 2012 ) [ arxiv:1107.5830 [ hep - ph ] ] .
a. j. barr , phys .
b * 596 * , 205 ( 2004 ) [ hep - ph/0405052 ] .
a. datta , k. kong and k. t. matchev , phys .
d * 72 * , 096006 ( 2005 ) [ erratum - ibid .
d * 72 * , 119901 ( 2005 ) ] [ hep - ph/0509246 ] .
j. m. smillie and b. r. webber , jhep * 0510 * , 069 ( 2005 ) [ hep - ph/0507170 ] .
l. -t .
wang and i. yavin , jhep * 0704 * , 032 ( 2007 ) [ hep - ph/0605296 ] .
m. burns , k. kong , k. t. matchev and m. park , jhep * 0810 * , 081 ( 2008 ) [ arxiv:0808.2472 [ hep - ph ] ] .
o. gedalia , s. j. lee and g. perez , phys .
d * 80 * , 035012 ( 2009 ) [ arxiv:0901.4438 [ hep - ph ] ] .
y. bai and h. -c .
cheng , jhep * 1106 * , 021 ( 2011 ) [ arxiv:1012.1863 [ hep - ph ] ] .
i. f. ginzburg , pos qfthep * 2010 * , 028 ( 2010 ) [ arxiv:1010.5579 [ hep - ph ] ] .
p. meade and m. reece , phys .
d * 74 * , 015010 ( 2006 ) [ hep - ph/0601124 ] .
m. r. buckley , h. murayama , w. klemm and v. rentala , phys .
d * 78 * , 014028 ( 2008 ) [ arxiv:0711.0364 [ hep - ph ] ] .
w. s. cho , k. choi , y. g. kim and c. b. park , phys .
d * 79 * , 031701 ( 2009 ) [ arxiv:0810.4853 [ hep - ph ] ] .
m. r. buckley , s. y. choi , k. mawatari and h. murayama , phys .
b * 672 * , 275 ( 2009 ) [ arxiv:0811.3030 [ hep - ph ] ] .
w. ehrenfeld , a. freitas , a. landwehr and d. wyler , jhep * 0907 * , 056 ( 2009 ) [ arxiv:0904.1293 [ hep - ph ] ] . c. -y . chen and a. freitas ,
jhep * 1201 * , 124 ( 2012 ) [ arxiv:1110.6192 [ hep - ph ] ]
. i. f. ginzburg , phys .
* 9 * , 678 ( 2012 ) .
l. edelhauser , k. t. matchev and m. park , jhep * 1211 * , 006 ( 2012 ) [ arxiv:1205.2054 [ hep - ph ] ] .
d. guadagnoli and c. b. park , arxiv:1308.2226 [ hep - ph ] .
t. han , i. -w .
kim and j. song , phys .
b * 693 * , 575 ( 2010 ) [ arxiv:0906.5009 [ hep - ph ] ] .
n. d. christensen , t. han , j. song and stefanus , arxiv:1404.6258 [ hep - ph ] .
n. d. christensen and c. duhr , comput .
commun . * 180 * , 1614 ( 2009 ) [ arxiv:0806.4194 [ hep - ph ] ] .
n. d. christensen , p. de aquino , c. degrande , c. duhr , b. fuks , m. herquet , f. maltoni and s. schumann , eur .
j. c * 71 * , 1541 ( 2011 ) [ arxiv:0906.2474 [ hep - ph ] ] .
t. tsukamoto , k. fujii , h. murayama , m. yamaguchi and y. okada , phys .
d * 51 * , 3153 ( 1995 ) .
s. y. choi , k. hagiwara , h. -u .
martyn , k. mawatari and p. m. zerwas , eur .
j. c * 51 * , 753 ( 2007 ) [ hep - ph/0612301 ] . | we construct a new kinematical variable that is able to fully reconstruct the absolute value , and partially reconstruct the sign , of the angular distribution in the center of momentum system of a decaying particle in certain cases where the center of momentum system is only known up to a two - fold ambiguity . after making contact with drell - yan production at the large hadron collider , we apply this method to the pair - production of dark matter in association with two charged leptons at the international linear collider and show that for a small intermediate width , perfect agreement is found with the true angular distribution in the absence of initial state radiation . in the presence of initial state radiation , we find that the modification to the angular distributions is small for most angles and that different spin combination classes should still be distinguishable .
this enables us to determine the spin of the mother particle and the dark matter particle in certain cases .
the existence of dark matter has been well established through a combination of galactic rotation curves @xcite , weak and strong gravitational lensing @xcite , big bang nucleosynthesis @xcite , the cosmic microwave background @xcite and the bullet cluster @xcite . from these observations ,
we know that dark matter is electrically neutral , non - baryonic and composes roughly 83% of the matter and 23% of the energy of the universe .
however , these observations do not tell us the detailed properties of dark matter such as its mass , spin and how it interacts with visible matter . for that , we need to observe a dark matter particle ( dmp ) in the laboratory . because the standard model ( sm ) of particle physics does not contain dark - matter ( among other things ) it is a low - energy effective theory that fits inside a larger , more complete theory .
two prominent examples of these theories are the minimal supersymmetric extension of the sm ( mssm ) and the universal extra - dimension ( ued ) model . in the present context ,
one of the most important features of these models is the presence of a new parity symmetry with the consequence that the lightest parity - odd particle ( lpp ) is stable and ( if neutral ) a dark - matter candidate @xcite . in these theories ,
the lpp is a weakly interacting massive particle ( wimp ) and , so , can be pair produced at particle colliders , such as the large hadron collider ( lhc ) and the international linear collider ( ilc ) . to determine the spin of a dmp at a collider , ideally , we would like to boost into the center of momentum ( cm ) frame of its parent particle and histogram the angle of its decay with respect to the boost direction ( see figure [ fig : bldsystem ] ) .
we will call this the cm angular distribution , where @xmath0 is the angle of the decay product @xmath1 with respect to the boost direction in the @xmath2 cm system .
if the width of the parent particle is narrow , this distribution will correspond with linear combinations of squares of the wigner @xmath3-functions where @xmath4 and @xmath5 correspond with the spin and spin - component along the boost direction of the parent particle @xmath2 and @xmath6 corresponds with the difference of the helicities of the final state particles @xmath1 and @xmath7 ( see appendix [ sec : wigner d - functions ] for a brief discussion . ) the challenge for dark - matter particles is that they do not interact with particle detectors and are , thus , not measured .
therefore , since we do not know their momentum , we often can not reconstruct the cm system .
decay of @xmath2 into @xmath1 and @xmath7 where @xmath1 is an observed sm - particle , @xmath7 is the missing dark - matter particle and @xmath2 is the parent of this decay . ] in this paper , we introduce a new kinematical variable that is able to fully reconstruct the absolute value of the cm angular distribution unambiguously and its sign up to a two - fold ambiguity even in some cases where the cm system is not known .
this method is a generalization of that used to reconstruct the spin of a new charged resonance in drell - yan processes at the lhc @xcite .
our result is the following .
the absolute value of the cosine of this angle is given by @xmath8 where @xmath9 , @xmath10 and @xmath11 are the masses of the parent particle ( @xmath2 ) , the observed particle ( @xmath1 ) and the dmp ( @xmath7 ) , @xmath12 is the component of the observed momentum of the visible particle ( @xmath1 ) that is transverse to the momentum of @xmath2 in the lab - frame , and @xmath13 the two possible signs are given by @xmath14 and @xmath15 which have been labeled by @xmath16 and @xmath17 , to be explained below and @xmath18 is the energy of the observed particle ( @xmath1 ) in the lab frame .
furthermore , @xmath12 can be expressed purely in terms of known quantities and the energy of @xmath7 in the lab frame as @xmath19 where @xmath20 the requirements for this method are that : the masses of @xmath2 , @xmath1 and @xmath7 must be known ; the full momentum of @xmath1 must be known ; the width of @xmath2 must be narrow ; @xmath2 , @xmath1 and @xmath7 must all be on - shell ; and either @xmath12 or @xmath21 must be known . under these circumstances ,
even if the b cm system can not be reconstructed , the cm angular distribution can be calculated , up to the sign ambiguity outlined above . in this paper
, we will describe two scenarios where these requirements are satisfied .
the first is in the discovery of a new resonance in charged drell - yan production of a charged lepton and a neutrino at the lhc .
we will summarize this scenario and refer to @xcite for further details .
the second is in the antler production of two charged leptons and two dark matter particles at the ilc , which we will describe in detail in this paper . before moving on
, we give a brief summary of other methods to measure the spin of dark matter .
an analysis of the spin - correlation in various cascade decay chains has shown that in many cases the resulting distributions were sufficient to determine the spin @xcite .
it has also been found that in certain cases the production cross - section varies with spin @xcite . additionally , the shapes of some other distributions have a dependence on the spin @xcite .
in addition to these methods , our method has the benefit of reconstructing the actual cm angular distribution even when the cm system can not be reconstructed , in many cases .
the rest of this paper is organized as follows . in section [ sec : derivation ] , we derive these kinematical variables in detail . in section
[ sec : dylhc ] , we summarize the application of these methods to charged drell - yan production and consider the effects of a finite width . in section [ sec : antlers ] , we describe how the application of our kinematical variables to antler processes at the ilc can be used to determine the spin of dark matter . in section [ sec : conclusion ] , we conclude . |
in december 2015 the atlas and cms collaborations have reported excesses in the search for resonances decaying into pairs of photons for diphoton invariant masses around 750 gev @xcite . in atlas
, excesses appeared in the two @xmath7 bins 710750 gev ( 14 events vs. 6.3 expected ) and 750 - 790 gev ( 9 events vs. 5.0 expected ) , with a local significance of @xmath8 ( assuming a large width of @xmath9 gev ; @xmath10 in the narrow width approximation ) . in cms , excesses appear in the @xmath7 bin 750770 gev for photons in the ebeb category ( 5 events vs. 1.9 expected ) and ebee category ( 6 events vs. 3.5 expected ) , but less in the bin 730 - 750 gev ( 4 events vs. 2.1 expected for photons in the ebeb category , 1 event vs. 4.0 expected for photons in the ebee category , considered as less sensitive ) .
the local significance of the excesses is @xmath11 for cms in the narrow width approximation .
the global significances of the signals of @xmath12 are not overwhelming and compatible with statistical fluctuations . still
, the fact that the region of invariant diphoton masses is very similar for atlas and cms has stirred quite some excitement resulting in a huge number of possible explanations .
( the number of proposed models exceeds the number of observed signal events . )
fits to the combined data should , in principle , also consider the informations from diphoton searches at 8 tev @xcite where a mild excess was observed by cms .
however , the extrapolation of signal cross sections from 8 to 13 tev depends on the assumed production mechanism @xcite . assuming the production of a resonance around 750 gev by gluon fusion ( ggf ) , combined fits to the signal cross sections at 13 tev are in the range 2 - 10 fb @xcite , with slightly better fits and a larger signal cross section assuming a larger width of 30 - 45 gev @xcite .
it is notoriously difficult to construct a consistent model for such a resonance `` @xmath13 '' : its production channel in proton proton collisions is typically assumed to be ggf through loops of colored particles
. if these are the quarks of the standard model ( sm ) , @xmath13 would decay into them leaving little branching fraction for @xmath13-decays into @xmath14 , which has to be generated by loop diagrams as well .
accordingly simple two higgs doublet ( or mssm ) extensions of the standard model , which could contain a resonance @xmath13 near 750 gev @xcite , require additional scalars or vector - like fermions whose loops generate the coupling of @xmath13 to gluons and/or @xmath14 ( unless @xmath0-parity is broken @xcite ) .
large yukawa couplings are required for a sufficiently large cross section , which risk to generate new hierarchy problems / landau singularities ( unless compositeness is invoked ) .
also in the next - to - minimal supersymmetric extension of the standard model ( nmssm ) it has been argued @xcite that additional vector - like quark superfields have to be introduced . in
@xcite a two - step decay cascade involving the two pseudoscalars of the nmssm with masses of about 750 gev and 850 gev has been proposed which requires , however , to tune the corresponding mixing angle close to 0 .
a different approach towards an explanation of the diphoton events is to consider that a single photon in the detector can represent a collimated bunch of photons ( typically two of them ) which originate from a single very light state , for instance a light pseudoscalar @xmath15 @xcite .
then the observed processes correspond to an initial resonance @xmath13 decaying into a pair @xmath16 , where @xmath15 must have a mass below @xmath17 gev for the resulting photons to be sufficiently collimated .
this scenario opens the possibility to explain the diphoton events in different models which can accomodate resonances @xmath13 and a light pseudoscalar @xmath15 . in this paper
we show that the simple @xmath18-invariant nmssm belongs to this class of models .
( this has also been observed in @xcite . ) in the nmssm ( see @xcite for reviews ) , two cp - even higgs states beyond the standard model - like higgs ( subsequently denoted as @xmath19 ) can play the role of a resonance @xmath13 . in terms of weak eigenstates ,
a singlet - like state @xmath20 can have a large coupling to a pair of mostly singlet - like pseudoscalars @xmath21 , originating from a cubic singlet self coupling @xmath22 in the superpotential ( see below ) .
however , a coupling to quarks or gluons inside protons has to be induced by a mixing of @xmath20 with one of the two su(2 ) doublet - like higgs states .
if this state is @xmath19 , the mixing reduces the couplings of @xmath19 to sm particles ( notably @xmath23 and @xmath24 ) and is severely constrained @xcite by the measured signal rates by atlas and cms @xcite .
an alternative is that @xmath20 mixes strongly with the other `` mssm''-like cp - even state @xmath25 .
then the physical eigenstates preferably both of them with masses near 750 gev can profit from an enhancement of the couplings of @xmath25 to @xmath26-quarks by @xmath27 , leading to sufficiently large signal cross sections into the @xmath28 ( and hence diphoton ) final state via associated production with @xmath26-quarks .
given the diphoton mass resolution of the detectors and the slightly preferred large width of the excess it is clear that two ( narrow ) cp - even states near 750 gev , mixtures of @xmath25 and @xmath20 , can also provide a good fit to the data .
( a similar scenario has been discussed in @xcite . ) for one of the benchmark points presented below ( bp1 ) the signal originates , however , from one cp - even state only , the other one being significantly heavier .
a light pseudoscalar can appear in the nmssm in the form of a pseudo - goldstone boson ( pgb ) .
a priori two global symmetries can lead to such pgbs : first , a peccei - quinn symmetry emerges in the limit @xmath29 @xcite .
however , @xmath30 is required for the couplings of the heavy higgs states to @xmath28 .
second , the scalar potential of the nmssm is invariant under an @xmath0-symmetry @xcite if the soft supersymmetry breaking trilinear couplings @xmath31 and @xmath32 vanish , leading to a pgb due to its spontaneous breakdown by the phenomenologically required vacuum expectation values .
we find indeed , that the interesting part of the parameter space of the nmssm corresponds to small values of @xmath31 and @xmath32 .
however , since the @xmath0-symmetry is broken by radiative corrections to the scalar potential involving the necessarily non - vanishing gaugino masses and trilinear couplings @xmath33 and @xmath34 , it helps only partially to explain a very light pseudoscalar @xmath21 .
still , it represents a `` go - theorem '' showing that a standard supersymmetric extension of the sm without additional vector - like quarks and/or leptons could explain the observed diphoton excess .
different assumptions on the mass of @xmath21 can be made . for one set of scenarios we assume @xmath35 mev , just below the @xmath36 threshold .
these scenarios lead to visibly displaced vertices from the @xmath37 decays . for a large value of the nmssm trilinear coupling @xmath38
, the signal can originate from a single higgs state near 750 gev .
for smaller values of @xmath22 , the signal can originate from two higgs states with masses near 750 gev . for another set of scenarios we assume @xmath39 mev , not far from the @xmath40 mass . for @xmath41 near 550 mev , @xmath21 mixes with the @xmath40 meson and inherits its decays into @xmath14 and @xmath42 ; the latter lead to photon - jets . for @xmath41 near 510 mev , constraints from searches for radiative @xmath43 decays into @xmath44 by cleo @xcite are alleviated , but estimates of the @xmath21 decay widths are more uncertain .
but in both cases the @xmath21 life time is short enough avoiding macroscopically displaced vertices , and two higgs states near 750 gev can generate a signal . in the next section we describe with the help of analytic approximations to the mass matrices ( including only the dominant radiative corrections ) which region in the parameter space of the nmssm can generate the diphoton events . in section 3
we discuss various constraints from low energy physics on light pseudoscalars , and discuss separately the different scenarios .
benchmark points are presented with the help of the public fortran code nmssmtools @xcite . in the final section 4
we summarize and discuss possible alternative signatures , which could help to distinguish different scenarios if the excess survives the next runs of the lhc .
we consider the cp - conserving @xmath18-invariant nmssm .
the superpotential of the higgs sector reads in terms of hatted superfields w_=_u_d + ^3 .
[ 2.1e ] once the real component of the singlet superfield @xmath45 develops a vacuum expectation value ( vev ) @xmath46 , the first term in @xmath47 generates an effective @xmath48 term = s .
the soft susy
breaking terms consist of mass terms for the gaugino , higgs and sfermion fields -l_&= & + , + -l _ & = & m_h_u^2 | h_u |^2 + m_h_d^2 | h_d |^2 + m_s^2 | s |^2 + m_q^2|q^2| + m_t^2|t_r^2| + & & + m_b^2|b_r^2| + m_l^2|l^2|+m_^2|_r^2| , as well as trilinear interactions between the sfermion and the higgs fields , including the singlet field -l _ & = & ( h_t a_t qh_u t_r^c + h_b a_b h_d q b_r^c + h_a_h_dl _
r^c + & & + a _
h_d s + a _
s^3 ) + .
the tree level scalar potential can be found in @xcite , from which the @xmath49 mass matrices in the cp - even and cp - odd sectors can be obtained .
once the soft higgs masses are expressed in terms of @xmath50 , @xmath27 and @xmath46 using the minimization equations of the potential , the mass matrices depend on the six parameters , , = , , a_a _ .
[ 2.5e ] initially , the cp - even mass matrix @xmath51 is obtained in the basis of the real components @xmath52 of the complex scalars @xmath53 after expanding around the vevs @xmath54 and @xmath46 .
it is convenient , however , to rotate @xmath51 by an angle @xmath55 in the doublet sector sector into @xmath56 in the basis @xmath57 : ^2_s = r ( ) ^2_s r^t ( ) , r ( ) = ( & & 0 + & - & 0 + 0 & 0 & 1 ) .
the advantage of this basis is that only the component @xmath58 of the higgs doublets acquires a vev @xmath59 and that , for typical parameter choices , it is nearly diagonal : @xmath58 has sm - like couplings to fermions and electroweak gauge bosons , the heavy doublet field @xmath60 is the cp - even partner of the mssm - like cp - odd state @xmath61 , while @xmath62 remains a pure singlet .
the mass matrix @xmath56 in the basis @xmath63 has the elements @xmath64 where @xmath65 and m_a^2=(a_+s ) is the mass squared of the mssm - like cp - odd state @xmath61 .
@xmath66 denotes the dominant radiative corrections due to top / stop loops , _ = ( ( ) + ( 1- ) ) [ 2.9e ] where @xmath67 and @xmath68 . as discussed in the introduction , we intend to describe the diphoton signal at @xmath69 gev by a mixture of the two states @xmath60 and @xmath62 .
then , both diagonal matrix elements @xmath70 and @xmath71 should have values close to @xmath72 . furthermore we will be interested in the @xmath0-symmetry limit @xmath73 .
this implies the relations ( for @xmath74 ) ^2_s,22~m_a^2 ~ ~ ^2~(750 ) ^2 [ 2.10e ] and ^2_s,33 ~(2s)^2 4()^2 ^2 ~(750 ) ^2 .
[ 2.11e ] the matrix element inducing @xmath75 mixing is given by ^2_s,23 ~2v , [ 2.12e ] and the matrix element inducing @xmath76 mixing by ^2_s,13 ~2v .
[ 2.13e ] next we turn to the cp - odd sector .
the @xmath49 cp - odd mass matrix contains always a goldstone boson which will be eaten by the @xmath24 boson .
the remaining cp - odd states are a singlet @xmath77 , and the `` mssm''-like su(2)-doublet @xmath61 . in the basis @xmath78 , in the @xmath0-symmetry limit @xmath73 , the cp - odd mass matrix is given by _
a^2 = ( s & -v 2 + -v 2 & ^2 2 ) .
[ 2.14e ] obviously @xmath79 has a vanishing eigenvalue @xmath80 , and is diagonalised by an angle @xmath81 with ( for @xmath82 ) .
[ 2.15e ] an important quantity will be the ( reduced ) coupling of @xmath21 to down quarks and leptons , which is obtained through the mixing of @xmath77 with @xmath61 .
since the reduced coupling of the mssm - like state @xmath61 is given by @xmath27 , one obtains x_d ~~ .
[ 2.16e ] radiative corrections to the tree level potential and hence to the cp - odd mass matrix include terms proportional to the electroweak gaugino masses @xmath83 and @xmath84 , and terms proportional to the soft susy breaking trilinear couplings @xmath33 and @xmath34 .
these corrections break the @xmath0-symmetry present for @xmath73 , which is expected since @xmath85 is not invariant under scale transformations .
hence , depending on the scale where @xmath85 is assumed to hold , @xmath21 is a pseudo - goldstone boson with a mass of typically a few gev . for @xmath32 small , but
@xmath86 one can obtain @xmath87 mev or @xmath88 mev as it will be assumed in the next section .
finally we note that , for the parameter region considered below , the dominant contribution to the coupling of @xmath21 to scalars originates from the quartic coupling @xmath89 .
after shifting @xmath62 by its vev @xmath46 one obtains g_s
a_1 a_1 ~ ^2 s .
[ 2.17e ] next we observe that eqs . and allow to express @xmath22 in terms of @xmath90 : from one finds 750 ~2s = 2 = where was used in the last step .
inserting @xmath91 gev one obtains ~1.1 x_d .
[ 2.19e ] in the next section , for the scenarios with @xmath39 mev , we will obtain upper bounds on @xmath90 from upper bounds for the @xmath92 from cleo @xcite .
these will thus imply upper bounds on @xmath22 according to .
on the other hand a large signal rate , generated by a mixture of the states @xmath60 and @xmath20 decaying into @xmath93 , requires @xmath94 to be as large as possible . accordingly
@xmath90 and @xmath22 should saturate corresponding upper bounds .
if the 750 gev signal is generated by a superposition of signals of two nearby physical states formed by the @xmath75 system , their mass splitting should not be too large , preferably of @xmath95 .
then the matrix element @xmath96 given in should be as small as possible . with @xmath22 already determined ,
this implies @xmath48 as small as possible , preferably close to the lower bound @xmath97 gev from the lep lower bound on higgsino - like charginos .
then requires that @xmath98 is relatively small .
( simultaneously , this avoids a strong push - down effect on the mass of the sm - like higgs boson from @xmath99 mixing , which is induced by the matrix element @xmath100 given in . ) finally the condition on @xmath101 fixes @xmath102 .
the remaining nmssm parameters in are @xmath31 and @xmath32 .
both @xmath0-symmetry breaking parameters have an impact on the mass of the pseudo - goldstone boson @xmath21 .
we find that one can chose small values of @xmath31 and @xmath32 such that @xmath41 assumes the desired value ; due to radiative corrections to the scalar potential the precise value of @xmath32 depends on the other @xmath0-symmetry breaking parameters @xmath83 , @xmath84 , @xmath33 and @xmath34 .
herewith all nmssm parameters are nearly uniquely determined .
as discussed in the introduction we will study scenarios with different values of the mass of a light pseudoscalar , denoted subsequently by @xmath41 .
constraints on such a light nmssm pseudoscalar with a mass below @xmath103 gev have been discussed previously in @xcite .
strong constraints originate from the mediation of fcncs . assuming minimal flavour violation , flavour violating couplings of @xmath21 still originate from susy loops involving stops , sbottoms and charginos and
depend on the corresponding masses and trilinear couplings like @xmath33 .
these contribute notably to @xmath104-physics observables like @xmath105 , @xmath106 and @xmath107 .
we have implemented the computation of these and many more @xmath104-physics observables and some @xmath108-physics observables in the code nmssmtools @xcite following the update in @xcite and checked that , for the scenario presented here , the constraints are satisfied due to the mostly singlet - like nature of @xmath21 and the relatively heavy susy spectrum . for @xmath41 near 210 mev , additional strong constraints originate from rare flavour changing processes @xmath109 .
( in @xcite it has been argued that the corresponding constraints exclude scenarios with @xmath110 mev , where the branching fraction of @xmath21 into @xmath111 is sizeable . )
we have verified the assertion in @xcite that , for suitable choices of soft susy breaking parameters , the coupling @xmath112 responsible for these processes ( see @xcite ) can be arbitrarily small .
light pseudoscalars have been searched for in radiative @xmath43 decays by cleo in @xcite ; these are also verified by nmssmtools_4.9.0 and satisfied by the benchmark points given below . due to the mostly singlet - like nature of @xmath21 , its contributions to the muon anomalous magnetic moment are negligibly small . however , for @xmath113 and assuming relatively light slepton masses of 300 gev , the scenarios below can reduce the discrepancy between the measured value and the standard model to an acceptable @xmath114 level .
further constraints stem from possible @xmath21 production in @xmath24 and @xmath19 decays .
the relevance of bounds on light pseudoscalars ( or axion - like particles ) from searches for @xmath115 at lep ( where a photon can correspond to a bunch of collimated photons ) has been investigated in @xcite .
these bounds constrain the loop - induced coupling @xmath116 .
this coupling is also constrained by the upper bound on @xmath117 @xcite .
we have checked that in our cases this coupling is about four orders of magnitude below the bounds derived from @xcite .
searches for @xmath118 have been undertaken by atlas using 4.9 fb@xmath119 of integrated luminosity at 7 tev c.m .
energy in @xcite for @xmath120 mev .
one can assume that the corresponding upper bound on @xmath121 applies to our scenario as well , which leads to @xmath122 .
if @xmath21 imitates a single photon , bounds on @xmath123 should be respected . in our scenarios
we require @xmath124 , hence these constraints are well satisfied .
notably this small branching fraction has no impact on the measured signal rates of @xmath19 into the other standard model channels , which agree well with the standard model predictions .
additional constraints depending on @xmath41 will be discussed in the corresponding subsections below . for a light @xmath21 , too light for hadronic final states ( @xmath41 below @xmath125 ) , the possible decays are into @xmath126 , @xmath111 and the loop induced decay into @xmath14 .
the couplings of @xmath21 to standard model fermions are obtained via mixing with @xmath61 as discussed in eqs . and in the previous section , and lead to a reduced coupling of @xmath21 to leptons @xmath127 , see .
these couplings determine also the partial width into @xmath14 . for a sizeable branching fraction into @xmath14 , the decay into @xmath126
must be kinematically forbidden . on the other hand , for @xmath128 mev
the remaining decays into @xmath111 and @xmath14 lead generically to a too small total width implying , for a boosted @xmath21 with an energy of about 375 gev , a decay length larger than the size of the detectors ( unless @xmath21 mixes strongly with @xmath3 as discussed in @xcite ) .
however , for @xmath41 very close to @xmath129 , the loop contribution of muons to the width @xmath130 reaches a maximum .
it is given by ( neglecting all other contributions ; see , e.g. , @xcite ) ( a_1)| _ = x_d^2 |a_1/2^a()|^2 with @xmath131 and , for @xmath132 , a_1/2^a(_f)=2 ^ -1 ^2 ; accordingly it increases with @xmath133 ( remaining finite for @xmath134 ) .
we find that , for @xmath41 near or slightly above 210 mev , the partial width @xmath135 dominated by the muon contribution is large enough to dominate the width @xmath136 leading to a @xmath137 .
the total @xmath21 width depends then essentially on @xmath90 related to @xmath22 via .
first we consider a scenario with a total width of @xmath138 gev , leading to a decay length of @xmath21 for an energy of 375 gev of about 2 m. given that the distance of the em calorimeter cells to the interaction point is larger than 1.3 m for the atlas and cms detectors ( depending on the angle @xmath40 ) , one can estimate that somewhat more than 60% of all pseudoscalars decay before the em calorimeter cells .
this scenario requires @xmath139 , in which case @xmath22 runs into a landau singularity at relatively low scales where the nmssm would require a uv completion ( e.g. gmsb ) .
then a single higgs state near 750 gev is able to generate a visible signal .
( the second higgs state is heavier near 1 tev and has a significantly smaller production cross section .
a scenario where a single higgs state near 750 gev is responsible for the signal and another higgs state is far below 750 gev is not possible : then the lighter state would generate a larger signal , which is excluded . ) for a large enough production cross section of the state near 750 gev from its coupling to @xmath26-quarks it must have a dominant @xmath60 ( mssm - like ) component .
still , for a large enough branching fraction into @xmath93 , the @xmath140 mixing angle in the heavy scalar higgs sector must not be too small and , notably , the coupling @xmath94 in must be large .
both of these conditions are satisfied for @xmath141 , which is required if a single state should generate a visible signal .
suitable values of @xmath98 , @xmath27 and @xmath48 for the desired masses and mixings are given by a benchmark point bp1 in table 1 .
( since the mass of the second heavy higgs state is near 1 tev and not near 750 gev , these values deviate somewhat from the ones obtained in the previous section . )
since @xmath27 is @xmath142 , the nmssm - specific uplift of the standard model like higgs mass at low @xmath27 is not available
. then the standard model like higgs mass of @xmath143 gev requires large radiative corrections as in the mssm . as stated above and discussed in @xcite ,
the squark masses and @xmath33 can be chosen such that flavour violating couplings of @xmath21 are suppressed . in order to generate simultaneously large enough radiative corrections to the standard model like higgs mass
, both parameters have to be relatively large in the multi - tev range .
possible numerical values are also indicated in table 1 .
the remaining nmssm specific parameters @xmath31 and @xmath32 are chosen small , such that the @xmath144 ( depending somewhat on @xmath31 ) is below @xmath145 , and @xmath41 sufficiently close to @xmath146 such that the total width of @xmath21 is large enough , i.e. that its decay length @xmath147 at 375 gev is small enough : for the bp1 in table 1 with @xmath148 mev we get @xmath149 gev and @xmath150 m , for which we estimate that @xmath151 of all @xmath21 decays take place before the em calorimeters .
( @xmath152 denotes the average distance to the calorimeter cells of @xmath153 m. ) for the production cross sections of the higgs state @xmath154 at 750 gev we find from sushi_1.5.0 @xcite ( at nnlo with mmht2014 pdfs ) @xmath155 fb , @xmath156 fb , and from nmssmtools we find @xmath157 with a total width of @xmath154 of @xmath158 gev .
together with a @xmath159 we obtain a signal rate of @xmath160 fb .
these latter values are shown in table 2 .
if we assume a slightly smaller value of @xmath161 mev , @xmath162 decreases to @xmath163 gev leading to @xmath164 m , reducing the acceptance to @xmath165 and hence the signal rate by @xmath166 .
scenarios with smaller values of @xmath22 are also possible . then , however , the reduced coupling @xmath90 of @xmath21 to leptons is smaller ( see ) , and the total @xmath21 width decreases . hence the decay length increases , and a smaller fraction of @xmath21 s decay before 2 m. this loss can be compensated for if two states @xmath154 and @xmath167 with large production cross sections and branching fractions into @xmath28 contribute to the signal .
the benchmark point bp2 is of this type , where we take @xmath168 , nearly ( but not quite ) small enough for the absence of a landau singularity below the gut scale . for @xmath169
mev the total @xmath21 width is @xmath170 gev , leading to @xmath171 m. we estimate that then only @xmath172 of all @xmath21 decays take place before the em calorimeter cells . on the other hand , two higgs states @xmath154 and @xmath167 with masses near 730 gev and 762 gev
contribute to the signal .
both are strong mixtures of the pure mssm - like and singlet - like states . for @xmath154 ,
the production cross section is @xmath173 fb , and @xmath174 . for @xmath167 ,
the production cross section is @xmath175 fb , and @xmath176 .
together with a @xmath177 we obtain a signal rate of @xmath178 fb , as shown in table 2 .
.parameters for the four benchmark points .
the soft susy breaking gaugino masses are @xmath179 gev ( 500 gev for bp1 ) , @xmath180 tev , @xmath181 tev , all squarks are assumed degenerate , and all slepton masses are 300 gev ( with vanishing trilinear couplings ) .
more digits ( for all parameters ) than shown here are necessary in order to reproduce @xmath41 given in the table 2 below , and in order to obtain @xmath112 suffiently small for bp1 and bp2 , see the text . [ cols="^,^,^,^,^",options="header " , ] the partial widths of a light pseudoscalar in this mass range can be estimated employing two complementary approaches . to begin with one can ask what one would obtain within the parton model , extrapolated into the nonperturbative domain of qcd . first , for a reduced coupling of @xmath21 to leptons @xmath182 as considered below , the partial width of @xmath21 into muons can still be computed reliably and is ( a_1^+^- ) ~510 ^ -11 .
the loop induced partial width of @xmath21 into @xmath14 is @xmath183 gev and hence negligibly small . at nlo
qcd the partial width of @xmath21 into strange quarks is about @xmath184 gev and the loop induced width into gluons of the same order as the width into @xmath126 .
these widths can only be rough estimates , however .
an alternative approach is to consider the case @xmath185 mev , where one can expect that @xmath21 mixes with the @xmath40 meson with a mass of 547.85 mev .
( the possible rle of @xmath40 for the decays of a light pseudoscalar has been indicated earlier in @xcite without quantitative statements , however . ) mixing with the @xmath3 meson of a lighter @xmath21 with @xmath186 mev has been considered in @xcite , where partial conservation of axial currents ( pcac ) or the sigma model for light mesons is employed in order to determine the off - diagonal element of the @xmath21-meson mass matrix ; the same formalism will be used here for @xmath187 mixing for @xmath185 mev .
first we discuss this latter case , where the results can be considered as more reliable .
only subsequently we turn to the case @xmath188 mev , motivated by the alleviation of constraints from radiative @xmath43 decays in this mass range , see below .
there , however , estimates of partial widths of @xmath21 are more speculative . for @xmath185 mev
, the relevant mass matrix of the @xmath187 system reads in the basis @xmath189 ( m_a_1 ^ 2 & m^2_a_1 + m^2_a_1 & m^2 _ ) . for a small mixing angle @xmath190 , ~1 , [ 3.5e ] the eigenstate @xmath191 contains a small @xmath40 component : @xmath192 . for the partial widths of @xmath191 one obtains then ( a_1x ) ( a_1x ) + ^2(x ) .
the dominant @xmath40 decays are @xcite [ 3.7e ] br()&~ & 39%,br(3 ^ 0)~33%,br(^+^-^0)~23% , + _ tot()&~ & 1.310 ^ -6 .
next we require that the @xmath40-induced decays into @xmath14 or @xmath42 of the eigenstate @xmath191 dominate its width into @xmath126 , since we ignore the unreliable widths of @xmath21 into strange quarks or gluons in this subsection .
( since the latter decays can also generate @xmath14 or @xmath42 final states , this assumption is conservative . )
this leads to ^2 > , 610 ^ -3 .
[ 3.8e ] in order to estimate the mixing matrix element @xmath193 above we use , following @xcite , pcac .
there one introduces the su(3)@xmath194 flavour currents @xmath195 where @xmath196 denote the su(3 ) generators . assuming that @xmath40 is a pure octet
, @xmath197 satisfies _
j^_a 8 = f_m_^2 with @xmath198 mev . at the quark level one has _
( |s^_5 s ) = -_j^_a 8 + _ j^_a 0 where @xmath199 is the ( anomalous ) u(1)@xmath200 current whose divergence involves the @xmath201 meson . using these relations
, one can re - write the coupling of @xmath21 to strange quarks in the lagrangian ( proportional to the corresponding yukawa coupling @xmath202 ) a_1 |s_5 s = - a_1 _ ( |s^_5 s ) = a_1 _ j^_a 8 + ...
= a_1 + ... [ 3.11e ] where we have dropped the terms @xmath203 . from one can read off m^2_a_1 = .
then the request becomes , again for @xmath182 and using , |m_a_1-m_| < 10 ^ -3 m_~0.5 .
[ 3.13e ] this estimate can be refined by including mixing with the @xmath201 meson , the anomalous u(1)@xmath200 current @xmath199 and the loop - induced coupling of @xmath21 to @xmath204 , where @xmath205 is the qcd field strength .
the additional contribution to @xmath193 leads to a replacement of the right hand side of by @xmath103 mev .
assuming such a small @xmath206 mass difference , the decay length of @xmath21 is below a mm , and its branching fractions are the ones of @xmath40 given in above . for @xmath207 ,
constraints from rare @xmath108 decays are no longer relevant . however , since @xmath21 has couplings to @xmath26-quarks @xmath208 , constraints from the search for the radiative decays @xmath209 by cleo in @xcite apply ( and are more relevant than searches for @xmath210 ) .
the @xmath211 can be obtained from the wilczek formula @xcite [ 3.14e ] & = & ( 1-)f , + br((1s)a_1 ) & ~ & 1.0310 ^ -4x_d^2 where @xmath212 and @xmath213 is a correction factor @xmath214 .
the upper bound of cleo @xcite on @xmath92 is @xmath215 at the 90% cl level , or @xmath216 at the 95% cl level . applying this bound to the @xmath211 , gives x_d 0.11 as used above .
from one finds that @xmath22 must then also be quite small , leading to relatively small branching fractions of the heavy higgs states @xmath154 and @xmath167 into @xmath28 .
hence both of these states should contribute to the signal .
next we consider the @xmath21 decays induced by its mixing with @xmath40 where @xmath40 decays as in .
the decays into @xmath217 give diphotons plus muons , but due to the escaping neutrinos this final state will not allow to reconstruct the masses of the original resonances near 750 gev .
in addition to the @xmath37 mode , the @xmath218 mode leads to photon jets .
the compatibility of such photon jets with a single photon signature in the detectors has been discussed in detail in @xcite . in particular , due to the enhanced probability for photon conversions into @xmath219 in the inner parts of the detectors , such scenarios can be distinguished from single photons ( or even diphotons ) once more events are available .
given the present small number of events ( which include the background ) one can just assume that such photon jets add to the signal .
adding optimistically both modes , about @xmath220 of all @xmath21 decays lead to di- or multi - photons .
the parameters , masses , branching fractions , production and signal cross sections of a corresponding benchmark point bp3 are shown in tables 1 and 2 .
if @xmath41 differs by a few tens of mev from the @xmath40 mass it becomes more difficult to estimate its decays ; its mixing angle with the on - shell @xmath40 meson using pcac as above becomes tiny .
its yukawa couplings to standard model fermions are obtained through mixing with the ( heavy ) mssm - like pseudoscalar @xmath61 . at the parton level and for @xmath221 ,
the relative couplings squared of @xmath21 are dominantly to @xmath222 ( @xmath223 ) , to @xmath224 via top quark loops ( @xmath225 ) , and to @xmath126 ( @xmath226 ) .
the hadronic or @xmath14 decays of @xmath21 can then be considered as being mediated by the cp - odd isospin and color singlet interpolating composite fields @xmath222 and @xmath204 . both are known components of the @xmath40 wave function in fock space , and the most reasonable assumption is that their hadronisation ( decays into physical hadrons and @xmath14 ) proceeds again with branching fractions similar to the ones of @xmath40 .
the partial width for the sum of these decays of @xmath21 is less clear , however .
it is relevant , since it competes with the width of @xmath21 into @xmath126 and determines consequently the branching fraction @xmath227 via the above interpolating fields relative to the @xmath228 .
since the widths for the above mentioned decays of @xmath40 into @xmath14 or pions are small ( being electromagnetic or suppressed by isospin ) , one must assume that the widths for the decays of @xmath21 via the above interpolating fields are also smaller than estimated from the couplings squared at the parton level as at the beginning of this section .
a quantitative statement is difficult , however , without a nonperturbative evaluation of the relevant matrix elements between physical states . on the other hand ,
the sum of the couplings squared of @xmath21 to @xmath222 or @xmath224 and hence the sum of the partial widths of @xmath21 into @xmath222 or @xmath224 ( for both of which @xmath40-like branching fractions are assumed ) is considerably larger than into @xmath126 : at nlo one has @xmath229 .
hence , reducing the sum of the partial widths of @xmath21 into @xmath222 or @xmath224 by a factor 1/10 leaves us still with a dominant @xmath230 . in the scenario where @xmath41 differs by a few tens of mev from the @xmath40 mass we will make the assumption that the reduction of the width of the decays @xmath231 is not too dramatic , i.e. the relevant branching fractions of @xmath21 can be parametrized as br(a_1,3 ^ 0,^0 ^+^- ) ~f_a br(,3 ^ 0,^0 ^+^- ) [ 3.16e ] where the factor @xmath232 is not too small @xmath233 .
let us have another look at the searches by cleo which were performed separately for the @xmath234 , @xmath235 and @xmath236 final states .
the windows for the invariant masses were chosen differently for different final states , @xmath237 mev and @xmath238 mev .
( @xmath7 is fitted to a double gaussian function centered at @xmath239 . )
no candidates were found in the ( background free ) @xmath42 and @xmath240 final states , but two events in @xmath240 with @xmath241 mev just below the @xmath242 window .
also a mild excess of events for @xmath243 mev is observed .
these events are not numerous enough to allow for the claim of a signal , but we conclude that the @xmath240 and @xmath14 final states do not lead to stronger upper limits on @xmath211 for @xmath244 mev than the limit on @xmath92 from the remaining @xmath42 final state . after translating the 90% cl upper limit from the latter final state into a 95% cl upper limit , we find from only the @xmath42 final state br((1s))0.33 3.810 ^ -6 [ 3.17e ] where 0.33 is the @xmath245 . for @xmath21 , assuming @xmath40-like decays , this upper bound becomes br((1s)a_1)0.33f_a 3.810 ^ -6 [ 3.18e ] combining and gives x_d 0.19/ [ 3.19e ] and , from , 0.21/ .
[ 3.20e ] since smaller @xmath232 alleviates the constraint on @xmath22 , we should study its impact on the signal rate .
first , the branching fractions of the heavy higgs states induced by the coupling @xmath94 in behave roughly like @xmath246 , i.e. smaller @xmath232 allows for larger branching fractions . on the other hand , by assumption ( see ) the branching fractions of @xmath21 into @xmath14 or @xmath42 are proportional to @xmath232 . hence the factors of @xmath232 cancel approximatively in the final signal rate .
( depending on the other parameters we found , however , that smaller @xmath232 can lead to a decrease of the signal rate if the heavy higgs branching fractions increase somewhat less than indicated above . ) for a rough estimate we have constructed a benchmark point bp4 with @xmath244 mev .
it has @xmath247 and satisfies the cleo constraints for @xmath248 .
its parameters are given in table 1 .
for the branching fractions of @xmath21 into @xmath14 or @xmath42 we assume @xmath249 which gives , using the corresponding branching fractions of @xmath40 , @xmath250 or @xmath251 . together with the production cross sections and branching fractions of the higgs states @xmath154 and @xmath167 in table 2 we obtain finally a signal cross section of @xmath252 fb .
we have shown that the excess of events in the diphoton final state near 750 gev observed by atlas and cms can be explained within a fairly standard supersymmetric extension of the standard model , the nmssm , without invoking new particles like additional vector - like quarks and/or leptons .
the signal cross sections are not very large , but may be sufficient to explain the observed excesses .
the corresponding processes differ , however , from what has been proposed in most of the literature up to now : except for the scenario bp1 ( with @xmath141 ) , two resonances nearby in mass which share the properties of the additional cp - even scalars of the nmssm are responsable for the signal cross section . their components proportional to the mssm - like scalar @xmath60 lead to enhanced couplings to @xmath26-quarks implying sizeable production cross sections via associated production with @xmath26-quarks , whereas their components proportional to the singlet - like scalar @xmath20 lead to sizeable branching fractions into two pseudoscalars @xmath93 .
these scenarios are not in tension with the upper limit from cms on the diphoton cross section obtained at the run i at 8 tev @xcite ; note that the @xmath253 cross sections increase somewhat faster with the c.m .
energy than @xmath254 .
we note that constraints from other decay modes of the scalars with masses of about 750 gev are satisfied : upper bounds on signal cross sections into other final states quark pairs , lepton pairs and electroweak gauge bosons are discussed in @xcite .
these bounds are obeyed given the relatively large branching fractions into @xmath93 and finally into di- or multi - photon final states of the heavy scalars in our scenario , which do not require excessive production cross sections .
four different scenarios have been discussed , which differ in the properties and masses of the light pseudoscalars @xmath21 and the heavy higgs states : \1 ) for the benchmark points bp1 and bp2 , the mass @xmath255 mev is just below twice the muon mass .
then the branching fraction of @xmath21 into diphotons is large enough for a satisfactory signal rate . in the case of the bp1 with @xmath141 , a single heavy higgs state ( still a mixture of the mssm - like and singlet - like states ) with a width of @xmath158
gev is sufficient for a signal . in the case of the bp2 with a more modest value for @xmath256 , two nearby heavy higgs states , both with a width of @xmath257 gev ,
are responsible for the signal . for both bp1 and bp2 ,
the dominant constraints from low energy experiments originate from @xmath108 decays involving loop - induced flavour changing vertices of @xmath21 ; it must be assumed that these are cancelled by suitable choices of the susy breaking parameters .
\2 ) for the benchmark points bp3 and bp4 it is assumed that @xmath21 shares its branching fractions with the @xmath40 meson . in the case of bp3 with @xmath258 mev this is guaranteed by @xmath187 mixing , estimated with the help of the pcac formalism . in the case of bp4 with @xmath244 mev , estimates of the @xmath21 partial widths are on less solid ground .
we assumed that the non - leptonic decays of @xmath21 proceed via @xmath222 or @xmath259 interpolating fields which , in turn , hadronise ( decay ) again similar to the @xmath40 meson .
we showed , however , that sizeable reductions of the corresponding partial widths with respect to the decays of @xmath21 into @xmath222 or @xmath224 by , e.g. , a factor 1/10 , would not invalidate this scenario . for both bp3 and bp4 two nearby heavy higgs states with widths of @xmath260 gev
are responsible for the signal . for this range of @xmath41
the dominant constraints from low energy experiments originate from searches for radiative @xmath4 decays into @xmath44 by cleo .
these lead to upper bounds on the coupling of @xmath21 to down - type quarks and leptons and , as we have shown , on @xmath22 .
interestingly , the four scenarios have different features which allow to distinguish them experimentally also from more `` conventional '' models : \1 ) for bp2 , bp3 and bp4 the signal originates from two resonances @xmath154 and @xmath167 close in mass , which can imitate a single wide resonance . of course , small variations of the parameters allow to vary the masses of @xmath154 and @xmath167 , the total signal rate , and to reshuffle the individual signal rates of @xmath154 and @xmath167 . with more events ( and depending on the actual mass difference ) the two states could possibly be resolved .
a particular feature of bp1 is that the single resonance near 750 gev responsible for the signal has another large branching fraction of @xmath261 into @xmath262 . with @xmath21 imitating a photon
, one obtains signals of the kind @xmath263 similar to the ones expected if a 750 gev resonance decays into @xmath14 via fermionic loops .
\2 ) the @xmath21 decays differ considerably for the benchmark points . for bp1 and
bp2 the decay lengths of @xmath21 are macroscopic leading to measurable displaced vertices if @xmath21 decays inside the calorimeters .
bp2 corresponds to an extreme case with a decay length of @xmath264 m , but a very large signal cross section ( before reducing it by the number of decays before the em calorimeters ) . for both bp1 and bp2
, @xmath21 decays into diphotons .
however , due to the displaced vertices it will not be straightforward to distinguish them from single photons via the number of converted photons @xcite .
moreover , @xmath21 has branching fractions of @xmath265 into @xmath219 leading to similar signatures as converted photons . for bp3 and bp4 the @xmath21 decay lengths are short , but
@xmath21 decays into diphotons or photon jets from @xmath42 .
the latter should lead to a very large proportion to `` converted photons '' ; additional potentially relevant observables like em shower shapes have also been discussed in @xcite .
finally , for bp3 and bp4 @xmath21 has branching fractions into muon pairs ( of @xmath266 at the parton level ) which could be used for alternative signals , once more events are obtained .
hence , if the excess of events continues , several observables can be used to verify / test / exclude the scenarios discussed here .
finally we recall that the origin of the light pseudoscalar @xmath21 in the nmssm is an approximate @xmath0-symmetry of the scalar potential , see the small values of @xmath32 and @xmath31 of the benchmark points . for bp1 ,
bp2 and bp3 , @xmath41 has to coincide accidentially with specific values @xmath267 or @xmath268 . for the bp4
@xmath41 is actually less constrained ( unless one intends to fit the events near @xmath244 mev observed by cleo as we did ) , but in this scenario the @xmath21 decays are less understood theoretically . although the approximate @xmath0-symmetry at the weak ( or susy ) scale is not preserved by radiative corrections , we content ourselves in the present paper with the mere fact that such a scenario would allow to explain the events .
work on an @xmath0-symmetric extension of the nmssm explaining a light pseudo - goldstone boson naturally is in progress .
the authors acknowledge the support of france grilles for providing cloud computing resources on the french national grid infrastructure and thank j. jaeckel and m. spira for helpful discussions , and notably f. domingo for communication on possible @xmath187 mixings .
u.e . acknowledges support from the european union initial training networks higgstools ( pitn - ga-2012 - 316704 ) , invisibles ( pitn - ga-2011 - 289442 ) , the erc advanced grant higgs@lhc , and from the grant h2020-msca - rise-2014 no .
645722 ( nonminimalhiggs ) .
the atlas and cms collaborations , `` measurements of the higgs boson production and decay rates and constraints on its couplings from a combined atlas and cms analysis of the lhc pp collision data at sqrt s = 7 and 8 tev , '' atlas - conf-2015 - 044 , cms - pas - hig-15 - 002 .
atlas collaboration , `` search for a higgs boson decaying to four photons through light cp - odd scalar coupling using 4.9 fb@xmath119 of 7 tev pp collision data taken with atlas detector at the lhc '' , atlas - conf-2012 - 079 . | the excess of events in the diphoton final state near 750 gev observed by atlas and cms can be explained within the nmssm near the @xmath0-symmetry limit .
both scalars beyond the standard model higgs boson have masses near 750 gev , mix strongly , and share sizeable production cross sections in association with b - quarks as well as branching fractions into a pair of very light pseudoscalars .
pseudoscalars with a mass of @xmath1 mev decay into collimated diphotons , whereas pseudoscalars with a mass of @xmath2 mev can decay either into collimated diphotons or into three @xmath3 resulting in collimated photon jets .
various such scenarios are discussed ; the dominant constraints on the latter scenario originate from bounds on radiative @xmath4 decays , but they allow for a signal cross section up to 6.7 fb .
lpt orsay 16 - 07 + lupm:16 - 004 + * ulrich ellwanger@xmath5 and cyril hugonie@xmath6 * + |
thurston ( * ? ? ?
* chapter 5 ) showed that a holonomy representation @xmath0 of the group of a hyperbolic knot @xmath1 in @xmath2 can be deformed into a one - parameter family @xmath3 of representations to give a corresponding one - parameter family @xmath4 of singular complete hyperbolic manifolds , the hyperbolic _ cone - manifolds _ of a knot @xmath1 .
let @xmath5 be a meridian of @xmath1 .
kojima @xcite showed further that @xmath6 is totally determined by the action of @xmath7 which is a rotation of angle @xmath8 around the fixed axis of @xmath9 .
a point on @xmath1 of the cone - manifold @xmath6 is in the core of a neighborhood isometric to a cylinder made of an angle @xmath8 wedge by identifying the two boundaries .
the @xmath8 is called a _ cone - angle _ along @xmath1 . a point off
@xmath1 has a neighborhood isometric to a neighborhood in @xmath10 .
we consider the complete hyperbolic structure on the knot complement as the cone - manifold structure of cone - angle zero . as we mentioned ,
if we increase the cone - angle from zero and if we keep the angle small , we get a one - parameter family of hyperbolic cone - manifolds .
similarly , for a link @xmath1 having @xmath11 components , we can get an @xmath11-parameter family of hyperbolic cone - manifolds of a link @xmath1 .
in particular , for each two - bridge hyperbolic link , there exists an angle @xmath12 for each link @xmath1 such that @xmath6 is hyperbolic for @xmath13 , euclidean for @xmath14 , and spherical for @xmath15 $ ] @xcite .
explicit volume formulae for hyperbolic cone - manifolds of knots and links are known only for a few cases .
the volume formulae for hyperbolic cone - manifolds of the knot @xmath16
@xcite , the knot @xmath17
@xcite , the link @xmath18 @xcite , the link @xmath19 @xcite , and the link @xmath20 @xcite have been calculated . in
@xcite a method of calculating the volumes of two - bridge knot cone - manifolds were introduced but without explicit formulae .
the main purpose of the paper is to find explicit and efficient volume formula for hyperbolic twist knot cone - manifolds .
the following theorem gives the formula for @xmath21 for even integers @xmath5 .
for odd integers @xmath5 , we can replace @xmath21 by @xmath22 as explained in section [ sec : twist ] .
so , the following theorem actually covers all possible hyperbolic twist knots .
but for the volume formula , since the knot @xmath23 has to be hyperbolic , we exclude the case when @xmath24 .
[ thm : main ] let @xmath23 be a hyperbolic twist knot .
let @xmath25 , @xmath26 be the hyperbolic cone - manifold with underlying space @xmath27 and with singular set @xmath23 of cone - angle @xmath8
. then the volume of @xmath25 is given by the following formula @xmath28 where for @xmath29 , @xmath30 with @xmath31 is a zero of the complex distance polynomial @xmath32 which is given recursively by @xmath33 with initial conditions @xmath34 where @xmath35 .
reflection ] a knot @xmath1 is a twist knot if @xmath1 has a regular two - dimensional projection of the form in figure [ fig : t2n ] .
for example , figure [ fig : knot ] is knot @xmath36 .
@xmath1 has 2 right - handed horizontal crossings and @xmath5 right - handed vertical crossings .
we will denote it by @xmath21 .
note that @xmath21 and its mirror image have the same fundamental group and hence have the same fundamental domain up to isometry in @xmath10 .
it follows that @xmath37 and its mirror image have the same fundamental set up to isometry in @xmath10 and have the same volume .
so , we will make no distinction between @xmath21 and its mirror image because we are calculating volumes . since the mirror image of @xmath21 is equivalent to @xmath22 , when @xmath5 is odd we will think @xmath22 as @xmath21 .
hence a twist knot can be represented by @xmath23 for some integer @xmath11 .
let us denote by @xmath38 the exterior of @xmath39 in @xmath27 . in ( * ?
* proposition 1 ) , the fundamental group of two - bridge knots is presented .
we will use the fundamental group of @xmath40 in @xcite . in @xcite ,
the fundamental group of @xmath40 is calculated with 2 left - handed horizontal crossings as positive crossings instead of two right - handed horizontal crossings .
the following proposition is tailored to our purpose .
[ thm : fundamentalgroup ] @xmath41 where @xmath42 .
we remark here that @xmath43 of proposition [ thm : fundamentalgroup ] is the meridian which winds around the bottom arc of the twist knot in figure [ fig : t2n ] and @xmath44 is the one that does the top arc as in figure [ fig : t2n ] .
let @xmath45 . given a set of generators , @xmath46 , of the fundamental group for @xmath47 , we define a set @xmath48 to be the set of all points @xmath49 , where @xmath50 is a representaion of @xmath47 into @xmath51 . since the defining relation of @xmath47 gives the defining equation of @xmath52 @xcite , @xmath52 is an affine algebraic set in @xmath53 .
@xmath52 is well - defined up to isomorphisms which arise from changing the set of generators .
we say elements in @xmath54 which differ by conjugations in @xmath51 are _
equivalent_. we use two coordinates to give the structure of the affine algebraic set to @xmath52 .
equivalently , for some @xmath55 , we consider both @xmath50 and @xmath56 : for the complex distance polynomial , we use for the coordinates @xmath57,\ ] ] @xmath58,\ ] ] and for the a - polynomial , @xmath59 \text { , } \ \ \ \eta^{\prime}(t)=\left[\begin{array}{cc } m & 0 \\
t & m^{-1 } \end{array } \right ] .
\end{array}\ ] ] since we are interested in the excellent component ( the geometric component ) of @xmath60 , in this subsection we set @xmath61 . given the fundamental group of a twist knot @xmath62 where @xmath42 , let @xmath63 and @xmath64 .
then the trace of @xmath65 and the trace of @xmath66 are both @xmath67 .
let @xmath68 be the complex distance between the axes of @xmath65 and @xmath66 in the hyperbolic space @xmath69 see ( @xcite , p. 68 ) for a formal definition of the complex distance ( width ) between oriented lines in @xmath69 the detailed formulae for calculation of @xmath70 can be found in the proof of theorem 4.3 .
[ lem : swc ] for @xmath71 which satisfies @xmath72 and @xmath73 , @xmath74 @xmath75 from the structure of the algebraic set of @xmath60 with coordinates @xmath76 and @xmath77 we have the defining equation of @xmath60 . by plugging in @xmath78 into @xmath79 of that equation and changing the variables to @xmath80 and @xmath81 , we have the following theorem .
[ thm : cpolynomial ] for @xmath80 , @xmath81 is a root of the following complex distance polynomial @xmath32 which is given recursively by @xmath82 with initial conditions @xmath83 note that @xmath84 , which gives the defining equations of @xmath60 , is equivalent to @xmath85 in @xmath86 by lemma [ lem : swc ] and @xmath85 in @xmath86 is equivalent to @xmath87 .
we may assume @xmath88 , \end{array}\ ] ] @xmath89 , \ \ \ t=\left[\begin{array}{cc } \cos \frac { \alpha}{2 } & i e^{-\frac{\rho}{2 } } \sin \frac { \alpha}{2 } \\
i e^{\frac{\rho}{2 } } \sin \frac { \alpha}{2 } & \cos \frac { \alpha}{2 } \end{array } \right ] , \end{array}\ ] ] and let @xmath90 .
then the equation , @xmath91 or @xmath92 gives the complex distance polynomial , where the third equality comes from the cayley - hamilton theorem . by direct computations ,
@xmath93 , @xmath94 , and @xmath95 have @xmath96 as a common factor .
hence , all of @xmath97 s have @xmath96 as a common factor .
actually , the common factor comes from the reducible representations . just as the a - polynomials
, we left the common factor out of our complex distance polynomials .
we divide @xmath97 by @xmath96 and denote @xmath98 by @xmath99 .
mathematica _ for the calculations .
let @xmath100 and @xmath101 , where @xmath102 is the word obtained by reversing @xmath103 . then @xmath104 and @xmath105 are longitudes which are null - homologus in @xmath40 .
we use @xmath105 for this subsection and use both @xmath104 and @xmath105 in section [ sec : pytha ] to keep the original form in @xcite and to keep the familar @xmath105 .
one can also deal with section [ sec : pytha ] with only @xmath105 or @xmath104 .
define @xmath106 to be a subset of @xmath107 such that @xmath108 and @xmath109 are upper triangular .
since every representation can be conjugated into this form , any element of @xmath54 is equivalent to an element of @xmath106 . by adding the equation stating that the bottom - left entry of the matrix corresponding to @xmath108 is zero ( the bottom - left entry of the matrix @xmath109 is already zero and the equation that the bottom - left entry of the matrix corresponding to @xmath108 is equal to zero is redundant in our setting )
, we have defining equations of @xmath106 and hence @xmath106 is an algebraic subset of @xmath54 .
define an _ eigenvalue map _
@xmath110 given by taking the top - left entries of @xmath108 , @xmath111 , and of @xmath109 , @xmath79 .
the closure of the image @xmath112 of an algebraic component @xmath113 of @xmath106 is an algebraic subset of @xmath114 .
if the closure of the image @xmath112 is a curve , there is a unique defining polynomial of this curve up to constant multiples .
the _ a - polynomial _ of the knot @xmath23 is defined by the product of all defining polynomials of image curves of @xmath106 .
the a - polynomial of a knot can be defined up to sign @xcite .
practically , if we let @xmath115 be the upper right entry of @xmath116 and @xmath117 be the upper left entry of @xmath108 , then the a - polynomial of the knot @xmath23 can be obtained by taking the resultant of @xmath118 and @xmath119 over @xmath44 , where the exponents @xmath120 and @xmath121 are chosen so that @xmath118 and @xmath119 become polynomials . in (
* theorem 1 ) , hoste and shanahan presented the a - polynomial of the twist knots .
let @xmath122 and @xmath123 . if we let @xmath124 and @xmath125 , then @xmath126 , @xmath127 and we have the following lemma .
[ lem : trace ] @xmath128 since @xmath129 we have @xmath130 [ def : longitude ] the _ complex length _ of the longitude @xmath104 or @xmath105 is the complex number @xmath131 modulo @xmath132 satisfying @xmath133 note that @xmath134 is the real length of the longitude of the cone - manifold @xmath25 .
we prepare and prove theorem [ thm : pytha ] for @xmath25 with @xmath135 . for @xmath136 , the same pythagrean theorem is obtained by replacing @xmath66 and @xmath137 with @xmath65 and @xmath138 .
we will use the oriented line matrix corresponding to a given matrix .
one can refer to ( * ? ? ?
* section v ) for oriented line matrices . denote by @xmath139 the line matrix corresponding to a matrix , @xmath140 , in @xmath86 . then @xmath141 . by sending common fixed points of @xmath66 and @xmath142 to @xmath143 and @xmath144 ,
we have @xmath145 , \ \ \ l_{\eta}=\left[\begin{array}{cc } e^{\frac{\gamma_{\alpha}}{2 } } & 0 \\ 0 & e^{-\frac{\gamma_{\alpha}}{2 } } \end{array } \right ] , \end{array}\ ] ] @xmath146,$ ] and the following line matrices @xmath147\\ & = \left[\begin{array}{cc } -i & 0 \\ 0 & i \end{array } \right ] , \end{aligned}\ ] ] @xmath148\\ & = \left[\begin{array}{cc } -i & 0 \\ 0 & i \end{array } \right ] , \end{aligned}\ ] ] which give the orientations of axes of @xmath66 and @xmath137 . ]
figure [ fig : polygon ] is the fundamental polyhedron for @xmath149 .
the double branched covering space of the polyhedron along @xmath150 and @xmath151 is the lens space @xmath152 .
the fundamental polyhedron for the hyperbolic cone - manifold of @xmath153 can be obtained from the fundamental polyhedron for @xmath149 by deforming the cone - angle continuously .
recall that a lambert quadrangle is a quadrangle with three right angles and one acute angle , not necessarily lying on a plane .
you can consult ( * ? ? ? * of section
vi ) or @xcite for the trigonometry of a lambert quadrangle or a right angled hexagon . let @xmath5 be the midpoint of @xmath154 .
then the quadrangle @xmath155 is a lambert quadrangle with acute angle @xmath156 , which can be considered as a right angled hexagon which is a generalized right angled triangle .
the six sides are @xmath157 . by applying the law of cosines to the hexagon
, we get the formula in the following theorem geometrically and the same argument works for all twist knots . hence , we call the following theorem pythagorean theorem
. now , we are ready to prove the following theorem which gives theorem [ thm : mpytha ] . recall that @xmath131 modulo @xmath132 is the _ complex length _ of the longitude @xmath104 or @xmath105 of @xmath25 .
( pythagorean theorem)[thm : pytha ] let @xmath25 be a hyperbolic cone - manifold and let @xmath68 be the complex distance between the oriented axes @xmath65 and @xmath66 .
then we have @xmath158 suppose @xmath135 . @xmath159
where the first equality comes from @xcite , the sixth equality comes from the cayley - hamilton theorem , and the seventh equality comes from lemma [ lem : trace ] .
let @xmath160 , @xmath161 , and @xmath162 . then @xmath163 , @xmath164 , and @xmath165 hence , @xmath166 which is equivalent to @xmath167 by solving for @xmath30 , we have @xmath168 by putting back @xmath169 , we have @xmath170 which is equivalent to @xmath158 pythagorean theorem [ thm : pytha ] gives the following theorem which relates the zeros of @xmath171 and the zeros of @xmath172 for @xmath173 , @xmath174 and @xmath175 .
[ thm : mpytha ] let @xmath175 and @xmath173 .
then the following formulae show that there is a one to one correspondence between the zeros of @xmath171 and the zeros of @xmath172 : @xmath176 with the same notation as in the proof of theorem [ thm : pytha ] , @xmath177 if we solve the above equation , @xmath178 for @xmath111 , we have @xmath179
we mention here that the proof can be done without referring to a - polynomial .
we identified @xmath111 with a root of @xmath180 because a - polynomial is rather well - known .
for @xmath135 and @xmath181 ( @xmath80 ) , @xmath180 and @xmath182 have @xmath183 component zeros , and for @xmath184 , @xmath185 component zeros . for each @xmath11
, there exists an angle @xmath12 such that @xmath25 is hyperbolic for @xmath13 , euclidean for @xmath14 , and spherical for @xmath15 $ ] @xcite . from the following equality [ equ : absl ] ,
when @xmath186 , which is equivalent to @xmath14 , @xmath187 .
hence , when @xmath8 increases from @xmath143 to @xmath188 , two complex numbers @xmath30 and @xmath189 approach to a same real number . in other words , @xmath190 has a muliple root and hence @xmath191 has a multiple root by theorem [ thm : mpytha ] .
denote by @xmath192 be the greatest common factor of the discriminant of @xmath193 over @xmath111 and the discriminant of @xmath194 over @xmath30
. then @xmath188 will be one of the zeros of @xmath192 . from theorem [ thm : mpytha ]
, we have following equality , @xmath195 for the volume , we can either choose @xmath196 or @xmath197 .
we choose @xmath111 with @xmath198 and hence we have @xmath31 by equality [ equ : absl ] . using the schlfli formula , we calculate the volume of @xmath199 for each component with @xmath198 and having one of the zeros @xmath188 of @xmath192 with @xmath12 on it . the component which gives the maximal volume is the excellent component @xcite . on the geometric component
we have the volume of a hyperbolic cone - manifold @xmath25 for @xmath26 : @xmath200 where the first equality comes from the schlfli formula for cone - manifolds ( theorem 3.20 of @xcite ) , the second equality comes from the fact that @xmath134 is the real length of the longitude of @xmath25 , the third equality comes from the fact that @xmath201 for @xmath202 by equality [ equ : absl ] since all the characters are real ( the proof of proposition 6.4 of @xcite ) for @xmath202 , and @xmath203 is a zero of the discriminant @xmath192 .
we note that the fundamental set of the two - bridge link orbifolds are constructed in @xcite .
we also note that the explicit formulae for the chern - simons invariant of the twist knot orbifolds are presented in @xcite and the a - polynomials of twist knots are obtained from the complex distance polynomials in @xcite .
table [ tab1 ] gives the approximate volume of @xmath204 for each n between @xmath205 and @xmath206 except the unknot and the torus knot and for each component with @xmath31 and having one of the zeros of @xmath192 with @xmath12 on it .
we used simpson s rule for the approximation with @xmath207 intervals from @xmath143 to @xmath188 . in that way
our approximate volume on the geometric component is the same as that of snappea up to four decimal points .
the geometric volume is written one more time on the rightmost column .
table [ table2 - 1 ] ( resp . table [ table2 - 2 ] ) gives the approximate volume of the hyperbolic twist knot cone - manifold , @xmath208 for @xmath11 between @xmath209 and @xmath206 ( resp .
for @xmath11 between @xmath205 and @xmath210 ) and for @xmath211 between @xmath212 and @xmath213 , and of its cyclic covering , @xmath214 .
we again used simpson s rule for the approximation with @xmath207 intervals from @xmath215 to @xmath188 .
mathematica _ for the calculations .
[ tab1 ] ll .volume of the hyperbolic twist knot cone - manifold , @xmath208 for @xmath11 between @xmath209 and @xmath206 and for @xmath211 between @xmath209 and @xmath213 , and of its cyclic covering , @xmath214 . [ cols="^,^,^",options="header " , ]
the authors would like to thank prof .
hyuk kim for his various helps and anonymous referees for their careful suggestions .
hugh m. hilden , mara teresa lozano , and jos mara montesinos - amilibia .
volumes and chern - simons invariants of cyclic coverings over rational knots . in _ topology and teichmller spaces (
katinkulta , 1995 ) _ , pages 3155 .
world sci .
publ . , river edge , nj , 1996 .
alexander d. mednykh .
trigonometric identities and geometrical inequalities for links and knots . in _ proceedings of the third asian mathematical conference , 2000 ( diliman ) _ , pages 352368 .
world sci .
publ . , river edge , nj , 2002 . | we calculate the volumes of the hyperbolic twist knot cone - manifolds using the schlfli formula .
even though general ideas for calculating the volumes of cone - manifolds are around , since there is no concrete calculation written , we present here the concrete calculations .
we express the length of the singular locus in terms of the distance between the two axes fixed by two generators . in this way
the calculation becomes easier than using the singular locus directly .
the volumes of the hyperbolic twist knot cone - manifolds simpler than stevedore s knot are known . as an application
, we give the volumes of the cyclic coverings over the hyperbolic twist knots . |
the stationary ( @xmath12 ) and non - interacting spin - susceptibility function , @xmath65 , is given in eq .
when the fermi - energy is at the hexagonal range and @xmath20 is close to one of nesting vectors , one of the eigenvalues of @xmath66 becomes much larger than the other two .
we denote the largest eigenvalue of @xmath66 as @xmath67 .
we found that @xmath67 is maximized at the nesting vectors of the fermi - surface .
now , we would like the examine the temperature dependence of @xmath67 . in order to do so , we set @xmath19 . the ratio of @xmath68 and @xmath69 as a function of the temperature appears in fig .
( [ fig : x_temp ] ) , where the temperature is measured in terms of @xmath70 .
clearly , the susceptibility increases as the temperature decreases .
however , it remains finite for all temperatures .
the critical interaction strength scales as @xmath71 , hence , its zero - temperature values are doubled at @xmath26 .
the values of @xmath72 at the other two nesting vectors are identical to fig .
( [ fig : x_temp ] ) due to symmetry considerations . for other @xmath20 vectors ,
the values of @xmath72 are different but the temperature dependence is similar .
the magnetization of a skyrmion - lattice is : @xmath73 where the summation is over the nesting vectors @xmath74 , @xmath75 are the eigenvectors of the spin - susceptibility , @xmath76 is the total amplitude of the skyrmion - lattice and @xmath77 is the phase of each sdw .
the mean - field hamiltonian of a strong topological insulator surface with eq .
( [ mag ] ) as an expectation value of @xmath78 is : @xmath79 where @xmath43 are the fourier components of @xmath44 , and @xmath80 is the non - interacting part which can be approximated by the fu model .
( [ h_mf ] ) can be solved numerically via bloch theorem .
that introduces a band structure and a set of eigenvectors , @xmath81 , at each point in the lattice brillouin zone .
the chern - number calculation can be done numerically by defining the projection matrix , @xmath82 , and using : @xmath83)\ ] ] in order to perform the calculation , we must set the parameters of the problem .
in particular , we must choose the exact form of the magnetization by setting the amplitude , @xmath76 , and the phases , @xmath84 , in eq .
( [ mag ] ) .
we found numerically that for any non - zero value of the amplitude @xmath76 , a gap is opened at the fermi - energy .
hence , any two states with two different positive ( or two different negative ) values of @xmath76 can be connected continuously without closing the gap .
consequently , they must share the same chern - number .
moreover , two states that differ only in the sign of @xmath76 are two time - reversal - partners and therefore their chern - numbers are equal in magnitude and opposite in sign .
hence , we conclude that the chern - number of the system can not depend on the magnitude of @xmath85 , and that the sign of @xmath76 affects only the sign of the chern - number . for that reason , we can choose a specific value of @xmath76 without losing the generality .
for the actual calculation , we chose @xmath86 and @xmath87 where @xmath88 is the appropriate critical interaction .
we set @xmath76 to be positive such that @xmath89 , where @xmath90 is the average magnetic energy per unit - cell : @xmath91 where the integration is over a lattice unit - cell and @xmath92 is a unit - cell area . the phases , @xmath84 , determine the origin of the skyrmion - lattice
. clearly , the spectrum and the chern - numbers are independent of these phases .
therefore , without loss of generality , we set all the phases to be zero .
after the determination of @xmath93 , @xmath25 and the exact form of @xmath40 , the numerical calculation of the chern - number is a straightforward task .
99 l. fu , c.l .
kane and e.j .
mele , phys .
lett . * 98 * , 106803 ( 2007 ) .
moore and l. balents , phys . rev .
b * 75 * , 121306 ( 2007 ) . l. fu and c. l. kane , phys .
* 100 * , 096407 ( 2008 ) .
m. levin , f. j. burnell , m. koch - janusz and a. stern , phys .
b * 84 * , 235145(2011 ) .
b. seradjeh , j. e. moore and m. franz , phys .
* 103 * , 066402(2009 ) .
j. maciejko , x. l. qi , a. karch and s. c. zhang , phys .
* 105 * , 246809(2010 ) .
y. baum and a. stern , phys .
b * 85 * , 121105(r)(2012 ) .
liang fu , phys .
lett . * 103 * , 266801(2009 ) .
jian - hua jiang and si wu , phys .
b * 83 * , 205124(2011 ) . y.l .
chen , j.g .
analytis , j.h .
chu , z.k .
liu , s.k .
mo , x.l .
qi , h.j . zhang , d.h .
lu , x. dai , z. fang , s.c .
zhang , i.r .
fisher , z. hussain , and z.x .
shen , science * 325 * , 178 ( 2009 ) .
d. hsieh , y. xia , d. qian , l. wray , j.h .
dil , f. meier , j. osterwalder , l. patthey , j.g .
checkelsky , n.p .
ong , a.v .
fedorov , h. lin , a. bansil , d. grauer , y.s .
hor , r.j .
cava , and m.z .
hasan , nature 460 , 1101 ( 2009 ) .
d. hsieh , y. xia , l. wray , d. qian , a. pal , j.h .
dil , j. osterwalder , f. meier , g. bihlmayer , c.l .
kane , y.s .
hor , r.j .
cava , and m.z .
hasan , science 323 , 919 ( 2009 ) .
t. h. r. skyrme , proc .
r. soc .
london a * 262 * , 233(1961 ) . s. l. sondhi , a. karlshede , s. a. kivelson and e. h. rezayi , phys .
b * 47 * , 16419(1993 ) .
s. m. girvin , in topological aspects of low dimensional systems , les houches lecture notes , edited by a. comtet , t. jolicoeur , s. ouvry and f. david ( springer - verlag , berlin / les editions de physique , les ulis , 2000);s .
m. girvin , cond - mat/9907002 ( unpublished ) .
l. brey , h. a. fertig , r. cote and a. h. macdonald , phys .
lett . * 75 * , 2562(1995 ) .
r. cote , a. h. macdonald , l. brey , h. a. fertig , s. m. girvin and h. t. c. stoof , phys .
* 78 * , 4825(1997 ) .
x. l. qi , t. l. hughes and s. c. zhang , phys .
b * 78 * , 195424(2008 ) .
d. j. thouless , m. kohmoto , p. nightingale and m. den nijs , phys .
lett . * 49 * , 405(1982 ) .
m. sitte , a. rosch , e. altman and l. fritz , phys .
* 108 * , 126807(2012 ) . | in this work we analyze the instability conditions for spin - density - waves ( sdw ) formation on the surface of strong topological insulators . we find that for a certain range of fermi - energies and strength of interactions the sdw state is favored compared to the unmagnetized and the uniform - magnetization states .
we also find that the sdw are of spiral nature and for a certain range of parameters a skyrmion - lattice may form on the surface .
we show that this phase may have a non trivial chern - number even in the absence of an external magnetic field .
the gapless surface states of strong topological insulators have drawn a great deal of attention over the past few years .
these metallic states arise from a topologically non - trivial band structure @xcite and are protected as long as time reversal symmetry is maintained .
this topological feature may lead to novel quantum states on the surface of these materials , such as majorana fermions and fractional excitations @xcite . in a previous work @xcite it was shown that for a strong enough electron - electron interaction , both of the short range and the coulomb types , the surface of a strong topological insulator is unstable to the formation of spontaneous uniform magnetization .
other works @xcite suggested that a spin - density - wave state ( sdw ) is likely to occur due to a strong fermi - surface nesting . in this work
we explore the conditions under which the sdw state is favored over the unmagnetized and the uniform - magnetization states .
moreover , we find that for a certain range of parameters , a skyrmion - lattice may form on the surface .
we also elaborate on the topological properties of this phase and claim that a network of one - dimensional chiral channels can be established on the surface .
we start by analyzing the spin - susceptibility function .
the non - interacting hamiltonian of a @xmath0 topological insulator surface can be approximated by the fu model @xcite : @xmath1 where @xmath2 , @xmath3 are the pauli matrices , @xmath4 is the electron velocity near the dirac point and @xmath5 is the warping parameter which originated from the cubic dresselhaus spin - orbit coupling of the bulk .
eq .
( [ eq : fu_h ] ) introduces an energy scale , @xmath6 . the fermi - surface , arising from eq .
( [ eq : fu_h ] ) , can be classified to three regions @xcite . a circular region for @xmath7 , a hexagonal region for @xmath8 and a snowflake region for @xmath9 . usually , the snowflake region coexists with the bulk states and therefore it is less interesting .
hence , we assume that the fermi - energy lies in the circular or the hexagonal regions .
we denote the energies and normalized eigenstates of eq .
( [ eq : fu_h ] ) as @xmath10 and @xmath11 .
the stationary ( @xmath12 ) and non - interacting spin - susceptibility function is given by : @xmath13 where @xmath14 is the fermi - dirac function at temperature @xmath15 with fermi - energy @xmath16 and @xmath17 .
the eigenvalues of @xmath18 , for @xmath19 , as a function of the fermi - energy are presented in fig .
( [ fig1 ] ) . clearly , as the fermi energy increases , one of the eigenvalues becomes much larger than the other two .
the maximal eigenvalue as a function of @xmath20 , for a fixed fermi - energy in the hexagonal region , is presented in fig .
( [ fig2 ] ) .
the susceptibility is maximized at the nesting vectors of the fermi - surface in the hexagonal region . by analyzing the susceptibility as a function of the temperature ( see supplementary ) , we find that the susceptibility increases dramatically as the temperature decreases .
however , it remains finite for all temperatures .
this feature arises from the fact that the fermi - surface is not perfectly nested , otherwise we would expect the susceptibility to diverge at zero temperature .
the landau free - energy up to a second order , with the magnetization , @xmath21 , as an order parameter , is given by : @xmath22 without interactions , as shown in fig .
( [ fig1 ] ) , the eigenvalues of the susceptibility are non - negative .
hence , eq . ( [ eq : fl ] ) is minimized by @xmath23 for every @xmath20 . however , in the presence of a contact interaction , the susceptibility can be expressed in terms of the non - interacting susceptibility using the random phase approximation ( rpa ) , @xmath24^{\rho\nu}$ ] , where @xmath25 is the contact - interaction strength . since the non - interacting susceptibility is finite for all temperatures ,
we conclude that a critical interaction at which one of the eigenvalues of the susceptibility becomes negative exists for all temperatures . above that critical value
a second order phase transition to a sdw ground state occurs .
the zero - temperature critical interaction for a formation of a sdw ground state as a function of the fermi - energy , as calculated in the rpa , appears in fig .
( [ fig3 ] ) . the critical interaction ( mean - field ) to have a uniform magnetization ground state as calculated in @xcite also appears in fig .
( [ fig3 ] ) . at the circular region , where the fermi - surface nesting is negligible ,
the uniform state is favored .
as the fermi - energy increases to the hexagonal range , the nesting becomes stronger and the sdw state is favored compared to the uniform one .
we also find that the values of the critical interaction for a formation of a sdw ground state are quite sensitive to temperature .
the zero temperature values in fig .
( [ fig3 ] ) are doubled at @xmath26 . above the critical interaction
a non - uniform magnetization will occur on the surface . by denoting the normalized eigenvectors of the susceptibility as @xmath27 ,
the magnetization becomes @xmath28 since all the eigenvectors of the susceptibility contain both real and imaginary parts , each @xmath20 contributes to the magnetization a spiral sdw propagating in the @xmath20 direction .
the magnitude of the sdw is almost constant and its direction varies continuously in space .
the relative phase between each pair of sdw , @xmath29 , is physical .
only two of these relative phases are independent , and choosing them is equivalent to choosing the origin of the sdw .
this freedom will be manifested as goldstone modes ( phonons ) . near the phase transition ,
only the nesting vectors @xmath30 will contribute to the magnetization .
however , these sdw do not necessarily coexist .
the landau free - energy up to a fourth order is : @xmath31 are four - spin correlation functions .
clearly , by tuning the ratio between @xmath32 and @xmath33 , the system can change its phase from a phase where all the spin - spirals coexist , to a stripes phase where only one of the spin - spiral has a magnitude that is different from zero .
in general , @xmath32 and @xmath33 depend on the dimensionless parameter @xmath34 and on the dimensionless fermi - energy ( @xmath35 ) .
though the exact calculation of @xmath32 and @xmath33 is difficult , a numerical mean - field calculation suggests that both phases may occur for different combinations of these parameters . to the end of this paper
, we will focus on the situation where the ratio between @xmath32 and @xmath33 is such that the favored ground state is the coexisting phase .
the superposition of the non - coplanar spin - spirals form then a triangular skyrmion - lattice on the surface @xcite .
the components of the magnetization form a periodic structure with a triangular symmetry .
each unit cell of the triangular lattice holds a skyrmion with a unit pontryagin - number @xcite , @xmath36 , and a zero net magnetization . in the core of each skyrmion the magnetization points in the direction normal to the surface and it rotates continuously towards the opposite direction at the edges of the unit cell .
an illustration of the magnetization direction in a unit cell appears in fig .
( [ fig4 ] ) . clearly , the system chooses spontaneously one of the two time - reversal - breaking states which are characterized by opposite pontryagin - number , @xmath36 .
in the @xmath37 quantum hall state @xcite , the skyrmion - lattice phase breaks both the translation and the @xmath38 symmetry of the single - skyrmion state .
unlike the quantum hall case , in this case there is no rotational symmetry and therefore we expect that the only massless goldstone modes will be the standard phonons of a two dimensional triangular lattice . motivated by the rpa results
, we now discuss the mean - field ( mf ) theory of the skyrmion - lattice phase .
this mean - field treatment will allow us a better understanding of the skyrmion - lattice phase and its topological features .
we consider eq .
( [ eq : fu_h ] ) as the non - interacting part of the hamiltonian .
we also assume a non - vanishing expectation value of the magnetization : @xmath39 where @xmath40 describes the skyrmion - lattice that we found in the rpa , and @xmath41 are the nesting vectors of the fermi - surface in the hexagonal region . in the mean - field approximation ,
the interaction part of the hamiltonian becomes : @xmath42 where @xmath43 are the fourier components of @xmath44 and @xmath25 is the interaction strength of a contact interaction .
the full mean - field hamiltonian , including the kinetic part eq .
( [ eq : fu_h ] ) , can be solved now by using bloch theorem , to yield a band structure . to the end of this paper ,
we restrict the fermi - energy to be at the hexagonal range and we assume that the magnetization energy is much smaller than that scale .
the two continuum branches of the dirac - cone split into a periodic band structure fig .
( [ fig5](a ) ) .
we define the ground - state mf energy : @xmath45 where the summation of @xmath46 is over the first brillouin zone and the summation of @xmath47 is over the occupied bands . by minimizing eq .
( [ eq : m_f_e ] ) with respect to the magnetization , we find that a critical interaction exists , above which there is a non - trivial minimum to eq .
( [ eq : m_f_e ] ) .
this critical interaction as a function of the fermi - energy appears in fig .
( [ fig3 ] ) .
we notice that the mf values of the critical interaction agree quite well with the rpa results .
we also find the phase transition to the gapped phase to be a second order phase transition where the gap grows as @xmath48 . as shown in @xcite , in the known topological insulators such as bi@xmath49te@xmath50 and bi@xmath49se@xmath50 , these values of critical interaction strength are too strong for an instability , albeit not by a large factor , making the instability an issue that may be relevant for other topological insulators .
the spectrum of the surface splits into many bands which , in general , cross each other .
however , we found that the two bands near the dirac - point cross each other only at the dirac - point and are gapped from all the other bands .
the two bands near the dirac - point for representative values of the parameters are presented in fig .
( [ fig5](b ) ) . clearly ,
while near the dirac - point the original branches of the dirac - cone are almost unchanged , at the fermi - energy the change is dramatic .
a gap is opened at the fermi - energy and the surface spectrum becomes fully gapped .
moreover , the surface spectrum is not symmetric with respect to the dirac point , since the magnetization has a non - zero pontryagin - number ( winding ) which differentiates between states with different chirality .
we now turn to explore the topological properties of the surface . in the absence of interactions ,
the spectrum of the surface is gapless . by adding a mass term , which is equivalent to a uniform magnetization
, a gap is opened at the dirac - point and a half - integer hall conductivity arises @xcite . for an infinite dirac - cone
, this half - integer hall conductivity is precisely quantized .
however , in real systems the dirac bands have an energy cutoff and therefore the half - integer hall conductivity is not precisely quantized .
it deviates from the half - integer value by an amount which scales as the gap energy divided by the cutoff energy .
moreover , the gap at the dirac - point can be opened due to interactions or by applying an external magnetic field . in both cases
the gap that is opened is rather small ( @xmath51 ) which makes the observation of the half - integer hall conductivity a hard task , since experimentally , it is required to fine tune the fermi - energy to the narrow gap .
as opposed to this narrow gap , the hexagonal range is covering a wide energy window , both in the upper and in the lower branch . in the well known topological insulators
this energy window , @xmath52 , is of the order of @xmath53 . as shown in fig .
( [ fig3 ] ) , by tuning the fermi - energy to any value in the hexagonal range and for a strong enough interaction , a gap is opened at the fermi - energy and a quantum hall effect may be observed .
the spectrum of the skyrmion - lattice phase is always gapped at the fermi - energy and the lattice symmetry of this phase yields a brillouin zone and a wave - function of the form of a bloch wave - function .
this allows us to define a chern - number associated with each band : @xmath54)\ ] ] where , @xmath55 is the projector to the @xmath56 band .
we denote the two bands near the dirac - point as @xmath57 and @xmath58 , where @xmath57 is the lower band and @xmath58 is the upper band ( see fig .
( [ fig5 ] ) ) .
of course , there are many more bands above and below these two , both surface and bulk bands . however , the @xmath59 bands are gapped from all other bands and we will treat them as an isolated band inside the bulk gap .
using eq .
( [ eq : chern ] ) , we find that when the fermi - energy is at the hexagonal range the chern - number of this band is @xmath60 ( see supplementary ) .
this chern - number of the combined band is independent of the existence of a mass term that introduces an energy gap between them . in the presence of a mass term these two bands
are separated and their total chern - number is unchanged .
it is instructive to examine how transport on the surface of a thick slab depends on the fermi - energy when electron - electron interaction is strong enough for spontaneous magnetization to occur .
when the fermi - energy is tuned to the hexagonal range in the lower branch , a gap is opened beneath the @xmath57 band and the occupied bands below the fermi energy contribute a chern - number @xmath61 .
the @xmath62 signs refer to the two possible time - reversal - breaking states . in general
, the contributions to this @xmath47 come from the three dimensional bands within the bulk of the slab and the occupied bands on its surfaces .
however , since the breaking of time - reversal - symmetry is limited to the surface , the bulk bands do not contribute .
as the fermi - energy is increased , the chern - number does not change as long as the system is within the skyrmion phase , since the energy gap does not close . once the fermi - energy reaches the region where the magnetization becomes uniform , the gap shifts away from the fermi - energy to the dirac point , and the hall conductivity ceases to be quantized .
it becomes quantized again when the fermi - energy is at the dirac point , in an energy gap .
then , the hall conductivity of the entire slab is quantized to an integer , and that of each surface is approximately quantized to a half - integer .
finally , when the fermi - energy is increased further , it passes through a gapless region in the circular range and gets to the hexagonal range on the positive side of the spectrum .
again , the system chooses spontaneously between the two time - reversal - breaking states and the occupied bands will contribute a chern - number of @xmath63 .
the particular value of @xmath47 is not universal , and would depend on details of the band structure .
we can , however , generally say that the difference in hall conductivity between the fermi - energy lying in the lower and upper hexagonal ranges is an odd integer .
these two cases are illustrated in fig .
( [ fig8 ] ) .
the analysis above points out to two types of situations where the hall conductivity on a surface may vary in space .
first , that may happen when time - reversal is broken in opposite ways in different regions . and second , that may happen when the fermi - energy varies in space , due to a variation between local density of electrons . in both cases , the domain walls must carry one - dimensional chiral edge channels .
notably , these chiral edge modes occur without an external magnetic field .
such one - dimensional edge channels on the surface of topological insulators may arise as a consequence of a magnetic field , as discussed in @xcite . here
we point out that in the presence of strong enough interactions , they may appear without an external field and without the need to fine tune the fermi - energy .
an illustration of this scenario is presented in fig .
( [ fig6],[fig7 ] ) .
the starting point is a topological insulator where the fermi - energy of all the surfaces is tuned to the hexagonal range at the lower branch . by tuning the fermi - energy of selected surfaces to the hexagonal range at the upper branch , one - dimensional edge channels will appear .
this tuning may be done by an external voltage . to summarize , we considered here the possibility that a non - uniform magnetization spontaneously appears on the surface of a three dimensional strong topological insulator due to interactions between surface electrons .
we assumed that the fermi - energy is such that the fermi - surface arising from a @xmath64 effective dirac hamiltonian is strongly nested .
we treated interactions within the mean - field and the random phase approximations .
we found that for a strong enough interaction the surface may be unstable to the formation of a skyrmion - lattice phase which is a result of a multiple sdw states .
we found the transition from the unmagnetized to skyrmion - lattice phase to be of second order as a function of interaction strength .
we also found that as long as the fermi - energy is located at the hexagonal range , the dependence of the critical interaction strength on fermi - energy is rather mild while the temperature dependence is rather strong .
we showed that even in the absence of an external magnetic field a non - trivial chern - number arises from the mean - field bloch wave function of the surface in the skyrmion - lattice phase and that this chern - number can be controlled by tuning the fermi - energy . using this feature
, we suggested that a network of one - dimensional chiral channels can be established on the surfaces of a strong topological insulator with a strong enough electron - electron interaction .
the authors thank the us - israel binational science foundation and the minerva foundation for financial support . |
the startup of the cern large hadron collider ( lhc ) will allow the exploration of the tev energy regime and the testing of the multitude of proposed theories of physics beyond the standard model .
many of these theories predict the existence of massive , neutral @xmath0-channel resonances @xcite . for some models of new neutral gauge bosons ( @xmath3 )
, the lhc is expected to have a discovery reach upwards of 5 tev with 100 fb@xmath4 of integrated luminosity @xcite .
this is a significant improvement over the current experimental limits for most models , which constrain @xmath3 masses to values greater than @xmath5 tev @xcite .
if a tev scale @xmath0-channel resonance were discovered , the immediate task would be to identify its origins .
many observables have been proposed to this end , primarily focused on the dilepton channel ( @xmath6 and @xmath7 ) , which would produce the cleanest and most easily measured signal for a non - leptophobic @xmath3 @xcite with the atlas @xcite and cms @xcite detectors .
the proposed measurements for the dilepton channel are the @xmath8 width , total cross section , forward - backward asymmetry ( @xmath2 ) @xcite , central - to - edge rapidity ratio @xcite , and a comprehensive analysis of all rapidity regions @xcite .
there are challenges associated with some of these measurements that we argue can be alleviated by using the pseudorapidity of the final state fermions .
in particular , we focus on two measurements : determining ( or at least constraining ) the spin of an @xmath0-channel resonance , and determining the forward - backward asymmetry of a @xmath3 . distinguishing whether the new resonance is a scalar , such as an r - parity violating sneutrino @xcite , a spin-2 boson , such as a kk graviton @xcite , or a spin-1 @xmath3 @xcite will be challenging and is typically determined through the study of the angular distribution in the centre - of - mass frame of the initial state quark and anti - quark ( c.m . )
the forward - backward asymmetry measurement at the lhc has to deal with the ambiguity in defining the forward direction due to the inability to unambiguously determine the direction of the initial state quark in a symmetric proton - proton collision .
presently , some solutions exist to deal with these challenges . to distinguish the spin of the resonance , a centre - edge asymmetry , @xmath1
, @xcite can be defined that is sensitive to the the angular distribution of the events .
the centre - edge asymmetry is a simple means of binning the events in the central and edge regions of @xmath9 , the c.m .
scattering angle , which will be weighted differently depending on the angular distribution .
this has the benefit of eliminating some of the systematic uncertainties of a fit to the angular distribution .
however , the @xmath1 observable still relies on boosting the particle four - momentum from the lab frame to the c.m
. frame .
the forward - backward ambiguity in a symmetric @xmath10 collision can be resolved by exploiting the fact that the valence quarks have , on average , larger momentum than the sea anti - quarks .
the quark direction can then be identified with the boost direction of the dilepton system .
@xcite . restricting the measurement to those events that have a large boost ( i.e. : @xmath11 )
reduces the misidentification of the initial state quarks and antiquarks , resulting in greater than @xmath12 of dilepton events being correctly identified as being boosted by the quark @xcite .
both of these methods have been explored in great detail and remain the standard approach used in the literature @xcite .
both the @xmath1 and the @xmath2 measurements require analysis of the centre - of - mass ( c.m . )
angular distribution of the dilepton events - directly for @xmath1 , and when tagging forward or backward events in @xmath2 . in this note
we propose a simpler method of measuring these asymmetries without reconstructing the angular distributions .
specifically , we exploit the direct measurements of the lepton pseudorapidities to calculate the observables , rather than using derived quantities that may propagate uncertainties into the result
. the proposed methods also take advantage of the fact that differences in pseudorapidities are lorentz invariant quantities , so that all calculations can be performed using quantities measured in the lab frame .
the point of these methods is not to provide new phenomenological insight into the models , but rather to demonstrate how the use of final state pseudorapidities provides a simpler and cleaner means of obtaining the @xmath1 and @xmath2 values .
the dimuon signal is very clean and error propagation should not be a big issue .
the real power of this approach will be seen when applied to heavy quark final states @xcite . in the following sections , we give some calculational details which are followed by a description of our approaches to the centre - edge asymmetry and the forward - backward asymmetry . we conclude with some final comments .
the basic ingredients in our calculations are the cross sections , @xmath13 , where @xmath14 , @xmath15 or @xmath16 .
the cross section for @xmath14 is described by the drell - yan process with the addition of a @xmath8 @xcite .
analagous expressions for the spin-0 @xmath15 and spin-2 graviton are given in ref .
@xcite and @xcite , respectively .
we computed the cross sections using monte - carlo phase space integration with weighted events and imposed kinematic cuts to take into account detector acceptances , as described in the following sections . in our numerical results
we take @xmath17 , @xmath18 , @xmath19 gev , and @xmath20 gev @xcite .
we used the cteq6 m parton distribution functions @xcite and included a k - factor to account for nlo qcd corrections @xcite .
we neglected nnlo corrections , which are not numerically important to our results @xcite , and final state qed radiation effects , which are potentially important @xcite but require a detailed detector level simulation that is beyond the scope of the present analysis .
the @xmath8 widths only include decays to standard model fermions and include nlo qcd and electroweak radiative corrections @xcite . for the @xmath15 width , we take @xmath21 gev following ref .
expressions for the @xmath16 width can be found in ref . @xcite .
the parton level angular distributions , @xmath22 , of the spin-0 , -1 , -2 bosons , shown in fig .
[ fig : ang_z ] , are distinct enough that , in principle , such a measurement would uniquely identify the spin @xcite . however , these distributions are not directly accessible due to the convolution with the parton distributions of the protons , the boosting of measured lab frame quantities to the centre - of - mass frame , detector limitations and finite statistics , all of which will make the measurement challenging @xcite . the centre - edge asymmetry is almost entirely model - independent for spin-0 and spin-1 bosons .
for example , assuming the narrow width approximation for a @xmath3 , we find that @xmath23 , for some value @xmath24 that separates the centre and edge regions of @xmath25 , independent of the couplings to fermions .
spin-2 kk gravitons have contributions from @xmath26 and @xmath27 processes that have slightly different angular distributions , and the @xmath1 depends on the weighted contribution of each .
the specific model will have an effect on the expected statistical uncertainties , but this should not be significant to the measurement due to the low backgrounds associated with leptonic final states .
thus , with limited statistics , an @xmath1 measurement could have an advantage over a fit to the angular distribution . for a hadron collider ,
the centre - of - mass angle of the outgoing fermion is not directly measurable on an event by event basis due to the unknown values of the parton momentum fractions .
however , there exists a direct mapping between the c.m .
angular distribution and the difference in pseudorapidity of the final state lepton and anti - lepton , @xmath28 .
furthermore , it is straightforward to show that @xmath28 is a lorentz invariant quantity , so that measuring this quantity in the lab frame is equivalent to measuring it in the centre - of - mass frame : @xmath29 the normalized @xmath30 distributions for spin-0 , -1 , -2 resonances are shown in fig . [
fig : norm_de ] , where it is clear that they are distinct from one another .
one can therefore construct a new centre - edge asymmetry using the lab frame @xmath30 distribution in place of the c.m .
frame angular distribution . using the mapping given by eq .
( [ eq : de_ct ] ) , we define the centre - edge asymmetry : @xmath31 following osland , _ et al . _
@xcite , we take @xmath32 , which they find to be the `` optimal '' value , and translates to @xmath33 . experimentally , detector acceptance constrains the pseudorapidity of each fermion to @xmath34 , which limits @xmath35 for the distribution . as is common in the literature , we take an r - parity violating @xmath15 with @xmath36 as an example of a spin-0 resonance @xcite , and an rs graviton with @xmath37 for a spin-2 resonance @xcite .
current experimental limits and studies on model parameters are given in ref .
@xcite and @xcite . for the @xmath3 case we explored the @xmath38 models ( @xmath39 , @xmath40 and @xmath41 ) @xcite , the left - right symmetric model ( lrsm , @xmath42 ) @xcite , both the littlest higgs ( lhm , @xmath43 ) @xcite and simplest little higgs ( slhm ) @xcite models , and the sequential standard model ( ssm ) .
the spin-0 model , spin-2 and some @xmath3 models we study predict narrow resonances , such that including events within several widths of the peak will be impossible in practice due to detector resolution effects smearing the breit - wigner distribution . instead , we examine events within one dilepton invariant mass bin as defined in the atlas tdr @xcite , using @xmath44 gev for the 1.5 tev resonance as in ref . @xcite .
a more precise measurement could be obtained by including events from a wider invariant mass window , if a broader peak were to be observed . in table
[ table : ace ] we show the expected centre - edge asymmetry for a spin-0 , spin-1 and spin-2 resonance , analogous to the study performed by dvergsnes , _ et al . _
@xcite , assuming muon final states with @xmath45 detection efficiency @xcite . from table [
table : ace ] one sees that if a @xmath8 were observed , a @xmath16 or @xmath15 could be ruled out .
likewise , an @xmath1 measurement would strongly discriminate against the @xmath8 or @xmath15 hypothesis if a @xmath16 were observed .
however , the @xmath8 and @xmath16 hypothesis could only be ruled out at approximately @xmath46 if a @xmath15 signal was observed .
the primary limitation in distinguishing between the different possibilities is the low statistics for @xmath15 production , as shown in the table , which is due to the tight constraints on the allowed values of its couplings .
other hypothetical spin-0 resonances may not be as tightly constrained and could therefore be distinguished from a @xmath8 or @xmath16 with higher statistical significance . .
@xmath47 values with corresponding statistical uncertainties for 100 fb@xmath4 integrated luminosity , @xmath48 gev , @xmath49 , within one bin @xmath50 gev and @xmath51 tev .
also shown are the expected number of total events for each model assuming 100 fb@xmath4 integrated luminosity . [ cols="<,<,>,<,<,^",options="header " , ] [ table : ace ]
the forward - backward asymmetry is a well - established measurement for distinguishing between models of @xmath3 s @xcite . for @xmath52 collisions at the tevatron
, the proton direction provides an obvious choice to define the
forward " direction .
the choice of forward direction at the lhc is more subtle , and is conventionally defined as the direction of the @xmath3 rapidity , @xmath53 .
the @xmath8 rapidity is chosen because the parton distribution functions for the valence quarks peak at a higher momentum fraction than those of the anti - quarks , so the system has a higher probability of being boosted in the quark direction .
this observation is statistical in nature and is more likely to hold true for larger values of @xmath54 . for smaller values of @xmath54 ,
the momentum fractions of the quark and anti - quark are generally closer in magnitude , so that using @xmath54 in the low rapidity region is less likely to correctly identify the quark direction . a simpler method of defining a forward " or
backward " event uses pseudorapidity . as before , we define the quark direction to be that of the higher momentum parton , or equivalently the direction of the @xmath3 rapidity .
one can then show that a forward " event is one in which @xmath55 in the lab frame , and vice - versa for a backward " event . using these definitions for forward and backward
, one can define the forward - backward asymmetry : @xmath56 dy}{\displaystyle\int \left[f(y ) + b(y)\right ] dy } \label{eq : afb}\ ] ] where @xmath57 is the number of forward events and @xmath58 is the number of backward events for a given @xmath59 , the @xmath3 rapidity ( i.e. : @xmath53 ) . the @xmath60 distribution under this definition is clearly shown in fig .
[ fig : dsdy ] to be symmetric in @xmath3 rapidity .
this method of finding @xmath2 has the advantage of being very straightforward and clean .
it simply relies on counting events with @xmath55 and those with @xmath61 .
we note that a related technique is employed by the cdf collaboration @xcite for the @xmath62 @xmath2 in @xmath52 collisions at the tevatron . however , the natural choice of the quark direction in @xmath52 collisions at the tevatron in contrast to @xmath10 collisions at the lhc results in important differences between the methods . as in the conventional method for finding @xmath2 , for small values of @xmath54 ,
there is a higher probability to wrongly assume that the quark is the parton with the higher momentum fraction .
this results in incorrectly assigning the forward or backward direction and gives a small `` wrong '' contribution to the @xmath2 measurement . for this reason
, it has been suggested that the central region , @xmath63 , be excluded in the measurement @xcite .
however , the coupling dependency can still be determined without this constraint on @xmath54 @xcite .
another consideration for excluding the central region is that the number of events that remain after subtracting @xmath64 is small , as shown in fig .
[ fig : dsdy ] , while the total number of events in this region is large .
excluding the events in the central region would increase the magnitude of @xmath2 , potentially making models more distinguishable .
however , we found that increasing @xmath65 resulted in an increase in the relative uncertainty .
we therefore conclude that little is gained by excluding events with small @xmath53 , and suggest that the whole rapidity region be included to decrease uncertainty and further simplify the @xmath2 measurement . using this method
, we calculate @xmath2 for the @xmath38 models ( @xmath39,@xmath40,@xmath41 ) @xcite , the left right symmetric model @xcite , the littlest higgs model @xcite , the simplest little higgs model @xcite , and the sequential standard model .
the on - peak versus off - peak @xmath2 are shown in fig .
[ fig : afb_on - off ] , where on - peak includes events which satisfy @xmath66 and off - peak includes events which satisfy @xmath67 , similar to the cuts used by petriello and quackenbush @xcite .
we conclude with the important observation that it might also be possible to include some events in the forward regions of the calorimeter ( fcal ) using this technique @xcite .
while a muon signature appears as missing @xmath68 in the fcal , it may be possible to distinguish an electron from a jet in the fcal due to the differences in the showering .
the signal would require triggering off of a single , high @xmath69 electron in the @xmath34 region , with an electron - jet in the fcal .
determining the charge sign of the single electron would distinguish whether this is a forward " or
backward " tagged event .
it is not clear what the signal efficiency of this method is , as reducible backgrounds include @xmath70 and others that might have low rejection rates .
extending the rapidity range has the potential of increasing the statistics and remains an interesting possibility for further study .
in this paper we described an approach for discriminating between various spin hypotheses for a newly discovered @xmath0-channel resonance at the lhc using a centre - edge symmetry , @xmath47 , that is based on the difference of the rapidities of final state fermions .
we also described a simple way to measure the forward - backward asymmetry , @xmath2 , using the properties of pseudorapidity .
both of these measurements have an advantage over previous approaches as they rely solely on the measurement of pseudorapidity , a fairly basic quantity .
the new measurements require simple counting and should propagate fewer errors than previous approaches that rely on boosting the four momentum into the centre - of - mass frame in order to perform the analysis .
our approaches reproduce the results found in other analyses but via a more straightforward analysis . j. l. hewett and t. g. rizzo , phys .
rept . * 183 * , 193 ( 1989 ) .
langacker , arxiv:0801.1345 [ hep - ph ] .
t. g. rizzo , arxiv : hep - ph/0610104 .
a. leike , phys .
* 317 * , 143 ( 1999 ) .
m. cvetic and s. godfrey , arxiv : hep - ph/9504216 .
l. randall and r. sundrum , phys .
lett . * 83 * , 3370 ( 1999 ) [ arxiv : hep - ph/9905221 ]
. j. l. hewett , phys .
lett . * 82 * , 4765 ( 1999 ) [ arxiv : hep - ph/9811356 ] . h. davoudiasl , j. l. hewett and t. g. rizzo , phys .
d * 63 * , 075004 ( 2001 ) [ arxiv : hep - ph/0006041 ] . j. kalinowski , r. ruckl , h. spiesberger and p. m. zerwas , phys .
b * 406 * , 314 ( 1997 ) [ arxiv : hep - ph/9703436 ] .
j. kalinowski , r. ruckl , h. spiesberger and p. m. zerwas , phys .
b * 414 * , 297 ( 1997 ) [ arxiv : hep - ph/9708272 ] .
s. godfrey , phys .
d * 51 * , 1402 ( 1995 ) [ arxiv : hep - ph/9411237 ] ; in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed .
n. graf , _ in the proceedings of aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) , snowmass , colorado , 30 jun - 21 jul 2001 , pp p344 _ [ arxiv : hep - ph/0201093 ] . t. aaltonen _ et al .
_ [ cdf collaboration ] , phys . rev .
lett . * 99 * , 171802 ( 2007 ) [ arxiv:0707.2524 [ hep - ex ] ] . j. erler , p. langacker , s. munir and e. r. pena , arxiv:0906.2435 [ hep - ph ] .
r. s. chivukula and e. h. simmons , phys .
d * 66 * , 015006 ( 2002 ) [ arxiv : hep - ph/0205064 ] .
g. aad _ et al . _ [ atlas collaboration ] , jinst * 3 * , s08003 ( 2008 ) . ``
atlas : detector and physics performance technical design report .
volume 1 , '' cern - lhcc-99 - 14 `` atlas : detector and physics performance technical design report .
volume 2 , '' cern - lhcc-99 - 15 g. l. bayatian _ et al .
_ [ cms collaboration ] , j. phys .
g * 34 * ( 2007 ) 995 .
p. langacker , r. w. robinett and j. l. rosner , phys .
d * 30 * , 1470 ( 1984 ) . f. del aguila , m. cvetic and p. langacker , phys .
d * 48 * , 969 ( 1993 ) [ arxiv : hep - ph/9303299 ] . f. petriello and s. quackenbush , phys .
d * 77 * , 115004 ( 2008 ) [ arxiv:0801.4389 [ hep - ph ] ] .
y. li , f. petriello and s. quackenbush , arxiv:0906.4132 [ hep - ph ] .
e. w. dvergsnes , p. osland , a. a. pankov and n. paver , phys .
d * 69 * , 115001 ( 2004 ) [ arxiv : hep - ph/0401199 ] .
p. osland , a. a. pankov , n. paver and a. v. tsytrinov , phys .
d * 78 * , 035008 ( 2008 ) [ arxiv:0805.2734 [ hep - ph ] ] .
p. osland , a. a. pankov , a. v. tsytrinov and n. paver , phys .
d * 79 * , 115021 ( 2009 ) [ arxiv:0904.4857 [ hep - ph ] ] .
v. d. barger , w. y. keung and e. ma , phys .
d * 22 * , 727 ( 1980 ) ; r. w. robinett and j. l. rosner , phys . rev .
d * 25 * , 3036 ( 1982 ) [ erratum - ibid .
d * 27 * , 679 ( 1983 ) ] ; s. capstick and s. godfrey , phys . rev .
d * 37 * , 2466 ( 1988 ) . see also ref .
@xcite and references therein . c. amsler
_ et al . _
[ particle data group ] , phys .
b * 667 * , 1 ( 2008 ) .
j. pumplin _ et al .
_ , jhep * 0207 * , 012 ( 2002 )
. j. kubar - andre and f. e. paige , phys .
d * 19 * , 221 ( 1979 ) .
k. melnikov and f. petriello , phys .
d * 74 * , 114017 ( 2006 ) . c. anastasiou
_ et al . _ ,
d * 69 * , 094008 ( 2004 ) .
u. baur _ et al .
_ , phys .
d * 65 * , 033007 ( 2002 ) ; u. baur , s. keller and w. k. sakumoto , phys . rev .
d * 57 * , 199 ( 1998 ) ; u. baur and d. wackeroth , nucl .
proc . suppl .
* 116 * , 159 ( 2003 ) .
a. l. kataev , phys .
b * 287 * , 209 ( 1992 ) .
t. han , j. d. lykken and r. j. zhang , phys .
d * 59 * , 105006 ( 1999 ) [ arxiv : hep - ph/9811350 ] .
j. bijnens , p. eerola , m. maul , a. mansson and t. sjostrand , phys .
b * 503 * , 341 ( 2001 ) [ arxiv : hep - ph/0101316 ] .
h. davoudiasl , j. l. hewett and t. g. rizzo , phys .
* 84 * , 2080 ( 2000 ) [ arxiv : hep - ph/9909255 ] .
r. n. mohapatra , _ unification and supersymmetry .
the frontiers of quark - lepton physics _ ( springer , berlin , 1986 ) .
n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * , 034 ( 2002 ) [ arxiv : hep - ph/0206021 ] .
m. schmaltz , jhep * 0408 * , 056 ( 2004 ) [ arxiv : hep - ph/0407143 ] . t. aaltonen _ et al .
_ [ cdf collaboration ] , phys . rev .
* 101 * , 202001 ( 2008 ) [ arxiv:0806.2472 [ hep - ex ] ] .
see also g. strycker , d. amidei , m. tecchio , t. schwarz , r. erbacher , j. conway , cdf public note 9724 .
[ http://www-cdf.fnal.gov/physics/new/top/2009/tprop/afb/ ] | we study the use of final state particle pseudorapidity for measurements of @xmath0-channel resonances at the lhc . distinguishing the spin of an @xmath0-channel resonance
can , in principle , be accomplished using angular distributions in the centre - of - mass frame , possibly using a centre - edge asymmetry measurement , @xmath1 . in addition , forward - backward asymmetry measurements , @xmath2 , can be used to distinguish between models of extra neutral gauge bosons . in this note
we show how these measurements can be improved by using simple methods based on the pseudorapidity of the final state particles and present the expected results for @xmath2 and @xmath1 for several representative models . |
a stack is a sorting device that works by a sequence of push and pop operations .
this last - in , first - out machine was shown by knuth @xcite to sort a permutation if and only if that permutation avoids the pattern @xmath8 .
that is , if there are not three indices @xmath9 with @xmath10 , then it is possible to run @xmath11 through a stack and output the identity permutation .
the class of stack - sortable permutations is enumerated by the catalan numbers . in the language of permutation patterns ,
any downset of permutations in the permutation containment ordering is a _ class _ , and every class has a _ basis _ , which consists of the minimal permutations not in the class .
given that the basis for the class of permutations sortable by one stack contains only a single pattern of length three , considering two stacks in series is quite natural .
however , the problem becomes rather unwieldy . in the case of two stacks in series ,
murphy @xcite showed that class of sortable permutations has an infinite basis .
the enumeration of this class also appears to be difficult .
the best known bounds are given by albert , atkinson , and linton @xcite .
we note that to sort a permutation by two stacks in series , the push and pop operations are such that when an entry is popped out of the first stack , it is immediately pushed into the second stack . to get a better handle on this problem ,
many have considered different types of weaker sorting machines .
one such weaker machine is a stack in which the entries must increase when read from top to bottom .
atkinson , murphy , and rukuc @xcite found an optimal algorithm for sorting permutations which are sortable using two increasing stacks in series .
their left - greedy algorithm sorts all permutations sortable by this machine .
( note that to obtain the identity permutation , the last stack will be an increasing stack even without declaring this restriction . )
interestingly enough , the basis for these sortable permutations is still infinite , but the permutation class was found to be in bijection with the permutations that avoid @xmath12 as enumerated by bna @xcite .
both enumerations were found by using a bijection with @xmath13 trees .
one can analogously define a _ decreasing stack _ as a stack in which the entries must decrease when read from top to bottom .
smith @xcite considered sorting with a decreasing stack followed by an increasing stack , a machine called di ( we refer to the decreasing stack as d and the increasing stack as i ) .
we illustrate how a permutation can be sorted with the di sorting machine in figure [ di_example ] .
notice that this permutation contains the pattern @xmath14 , and as such can not be sorted by two increasing stacks in series .
[ cols="^,^,^,^ " , ] we begin by producing @xmath15 by appending a diagonal ( figure [ fig : graph_inverse](a ) ) .
we identify @xmath16 and @xmath17 based on the relative locations of the diagonal edges with respect to the north edges .
we construct @xmath18 by converting each diagonal edge to a north edge ( figure [ fig : graph_inverse](b ) ) .
next we find @xmath19 and construct @xmath20 by inserting ` e ` before the ` n ` at each of the locations in @xmath21 specified by @xmath22 .
note that @xmath23 , and with this we construct @xmath24 .
observe that @xmath25 is the adi word of 81736245 .
while the bijection obtained from algorithm [ adi_word_to_path ] can be used to show exactly which schrder path corresponds to a given di - sortable permutation , there are several properties that translate prominently between permutation and path .
[ number_rtlm ] if @xmath25 is the adi word of a di - sortable permutation @xmath11 , then @xmath28 if and only if @xmath29 is a right - to - left minimum .
furthermore , the number of right - to - left minima of a di - sortable permutation @xmath11 is one more than the number of diagonal steps in its corresponding schrder path .
suppose @xmath29 is a right - to - left minimum of @xmath11 with adi word @xmath25 .
there must be some @xmath30 $ ] that satisfies @xmath31 by theorem [ en_before_c ] .
assume by way of contradiction that @xmath32
. then @xmath33 and @xmath34 .
hence @xmath35 and @xmath36 , which is a contradiction to @xmath29 being a right - to - left minimum .
so @xmath28 .
now suppose that @xmath28 for some @xmath37 $ ] and let @xmath38 $ ] such that @xmath39
. then @xmath40 .
therefore @xmath41 , so @xmath42 .
hence @xmath43 .
so @xmath29 is a right - to - left minimum .
hence there is a one - to - one correspondence between consecutive sequences ` enc ` in @xmath25 and right - to - left minima in @xmath11 . and as seen in algorithm [ adi_word_to_path ] , these sequences give rise to one fewer diagonal in the corresponding schrder path of @xmath11 . [ catalan ] the di - sortable permutations with one right - to - left minimum , that is the di - sortable permutations that end in @xmath44 , are enumerated by the catalan numbers . in particular , the number of di - sortable permutations of length @xmath45 ending in @xmath44 is @xmath46 .
alternatively , recall that the basis for our di - sortable permutations is @xmath48 .
thus , the di - sortable permutations of length @xmath45 ending in @xmath44 are exactly the permutations whose first @xmath0 entries must avoid @xmath49 , which are known to be enumerated by catalan numbers , @xmath50 @xcite . for a given diagonal in a schrder path @xmath51 and prefix @xmath52 of @xmath51 terminating after the given diagonal ,
the we say the _ height _ of the diagonal is @xmath53 .
in other words , its height is the @xmath54-coordinate of the point in the path where the diagonal terminates .
[ diagonal_height ] the heights of the diagonals determine the values of the right - to - left minima . in particular , if the heights of the diagonals in the schrder path corresponding to the permutation @xmath11 are @xmath58 , the right - to - left minima of @xmath11 are @xmath59 .
suppose that @xmath51 has @xmath60 diagonals ( and hence @xmath11 has @xmath61 right - to - left minima ) , and let @xmath25 be the adi word of @xmath11 .
notice since @xmath62 counts the ` c`s in the first @xmath63 maximal substrings of ` c`s , we have @xmath62 counts @xmath64 . as such ,
the right - to - left minimum are exactly @xmath65 for @xmath66 .
( recall @xmath67 by convention . )
thus @xmath68 if and only if @xmath69 for some @xmath63 where @xmath66 .
let @xmath70 be locations in @xmath51 so that @xmath71 is the location of the @xmath72 ` d ` ; that is @xmath73 and @xmath74 . from step 3 in algorithm [ adi_word_to_path
] , we observe that the height of the @xmath72 diagonal is @xmath75 for each @xmath76 $ ] .
so the heights of the diagonals of @xmath51 are one less than the right - to - left minima of @xmath11 ( excluding 1 ) .
a permutation @xmath11 is said to be _ plus - decomposable _ if @xmath11 is the concatenation of two non - empty intervals @xmath79 and @xmath80 where the values of @xmath79 are less than those of @xmath80 .
further , if we rescale the entries of @xmath80 by subtracting the length of @xmath79 from each entry of @xmath80 to get a permutation @xmath81 , we denote @xmath82 .
if a permutation is not plus - decomposable , we say the permutation is _ plus - indecomposable_. the schrder path corresponding to a plus - decomposable di - sortable permutation @xmath82 is the schrder path for @xmath79 followed by a diagonal step ( on the main diagonal ) followed by the schrder path for @xmath81 .
furthermore , this means the schrder paths with diagonal steps on the main diagonal correspond exactly to the plus - decomposable di - sortable permutations .
consider the plus - decomposable permutation @xmath82 . by corollary [ greedy
] , we know algorithm [ sorting_algorithm ] will force the exit of all the entries of @xmath79 before any entry of @xmath80 enters the stacks .
thus the adi word for @xmath11 will be the adi word for @xmath79 followed by the adi word for @xmath81 .
when algorithm [ adi_word_to_path ] acts on the adi word for @xmath11 , it will convert the first portion of the adi word to the schrder path for @xmath79 , but followed by @xmath85 as this @xmath85 is no longer the last @xmath85 of the word . at the same time
, it converts the remaining portion of the adi word to the schrder path for @xmath81 .
[ ex : plusdecomp ] in figure [ fig : plusdecomp ] , we see that this schrder path for @xmath86 is the concatenation of the schrder paths for @xmath87 , @xmath8 , and @xmath44 ( a degenerate path ) , connected by diagonal edges .
aguiar and moreira @xcite showed that the number of schrder paths from @xmath2 to @xmath47 that do not have diagonal steps on the main diagonal equals the number of schrder paths from @xmath2 to @xmath47 that do have diagonal steps on the main diagonal as a particular case of a larger result .
that is , they showed the number of schrder paths that do not have diagonal steps on the main diagonal are enumerated by the small schrder numbers , which gives us the following corollary . 1
. are there any other properties of the di - sortable permutations that are easily seen by looking at the corresponding schrder paths ? 2 .
is there another bijection between the di - sortable permutations and the schrder paths that conserves more or different properties of the di - sortable permutations in the schrder paths ? | we consider a sorting machine consisting of two stacks in series where the first stack has the added restriction that entries in the stack must be in decreasing order from top to bottom .
the class of permutations sortable by this machine is known to be enumerated by the schrder numbers . in this paper
, we give a bijection between these sortable permutations of length @xmath0 and schrder paths of order @xmath1 : the lattice paths from @xmath2 to @xmath3 composed of east steps @xmath4 , north steps @xmath5 , and diagonal steps @xmath6 that travel weakly below the line @xmath7 . |
recently there has been much interest @xcite in the ads / cft approach @xcite to study the non - perturbative aspects of gauge theories at large @xmath1 and strong coupling with even applications to heavy ion collision experiments such as jet quenching @xcite . on the other hand ,
dense qcd is of much relevance to compact stars and current heavy ion collisions .
this subject has been intensely studied in the past decade following the observation that at asymptotic densities qcd matter may turn to a color superconductor due to asymptotic freedom @xcite . at very large @xmath1
there is evidence that the overhauser effect takes over with the formation of density waves @xcite .
indeed , at weak coupling the high degeneracy of the fermi surface causes quark - quark ( antiquark - antiquark ) pairing of the bcs kind with different color - flavor arrangements , while at large number of colors quark - antiquark pairing of the overhauser kind is favored . in strong coupling ,
first principle calculations are elusive and the two pairings may compete . to address the issue of strongly coupled qcd at large @xmath1 in dense matter
we explore in this paper the holographic principle . in recent years
, the gauge / string duality has provided a framework for addressing a number of problems in strong coupling where little is known from first principles in the continuum .
to do so , we consider a model for holographic qcd recently discussed by sakai and sugimoto . in the limit @xmath5 , chiral symmetry in qcd
is generated by immersing @xmath6 d8-@xmath7 into a d4 background in 5 dimensions where supersymmetry is broken by compactification ( kaluza - klein mechanism with @xmath8 scale ) .
the induced dbi action on d8 yields a 4 dimensional effective theory of pions and infinitly many vector mesons where the @xmath8 scale plays the role of an upper cutoff ( the analogue of the chiral scale @xmath9 ) . at large @xmath1 baryons are skyrmions . in this paper , we extend the analysis by sakai and sugimoto @xcite to finite baryon density .
although there have been many papers @xcite on finite chemical potential in the holographic approach , all involve the r - charge or isospin instead of fermion number .
the difficulty for the fermion number is that the @xmath10 charge for r - symmetry is already dual to the @xmath10 charge of ads reisner - nordstrom black hole charge .
there is no additional @xmath10 in bulk to modify the geometry of the ads black hole . in this paper
we follow the more traditional route of introducing the chemical potential via the nonperturbative induced dbi plus cs action through the external @xmath11 source in sakai and sugimoto chiral model .
specifically , we introduce the baryon chemical potential as @xmath12 where @xmath13 is the external vector field in the induced dbi plus cs action . while @xmath14 drops in the dbi part it contributes to the cs part through the wzw as it should .
a redefinition of the vector fields in the sakai - sugimoto model necessary to enforce vector dominance , causes a reshuffling of @xmath14 from the cs to the dbi action as we detail below , with important phenomenological consequences . in section 2
we review the emergence of the induced effective action on the probe d8 brane from the 5 dimensional d4-brane background , and we detail the contributions of the dbi action to fourth order in the vector fields , as well as the chern - simons part .
the expanded effective action enforces total vector dominance . in section 3
we introduce the baryon chemical potential and details the nature of the effective expansion at finite density . in section 4 ,
we extract the baryon number density , the pressure and the energy density of holographic dense matter . in section 5 , we discuss the pion and leading vector meson parameters and interactions in holographic dense matter .
our conclusions and suggestions are summarized in section 6 .
the _ generalized holographic principle _ states that at large @xmath1 maximally supersymmetric and conformal gauge theories in flat d - dimensions are dual to superstring / mtheory on pertinent ads spaces , i.e. for @xmath15 it is @xmath16 and for @xmath17 it is @xmath18 . to model qcd in @xmath15 dimensions ,
we start from @xmath17 superconformal theory , and compactify it twice .
first through @xmath19 with radius @xmath20 leading to @xmath21 susy theory .
second through another @xmath19 of radius @xmath22 with fermion antiperiodic boundary conditions to break susy . below the cutoff scale @xmath23 , the d=5 broken susy theory behaves as a d=4 gauge theory .
doubly compactified d=6 is the boundary of @xmath24 in which the m - theory reduces to type iia string theory . as a result m5 branes wrapping around the first @xmath19 with radius @xmath25 transmute to d4-branes .
@xmath1 copies of these branes yield near extremal black - hole background .
the metric , dilaton @xmath26 and the rr three - form field @xmath27 in a d4-brane background are given by @xmath28 here @xmath29 and @xmath30 is the compact variable on @xmath19 .
@xmath31 is the radial coordinate along @xmath32 , @xmath33 the 4-form and @xmath34 the volume of a unit @xmath35 surrounding the d-4 brane .
@xmath36 , where @xmath37 and @xmath38 are the string coupling and length respectively .
again , this background represents @xmath1 d-4 branes wrapped on @xmath19 .
let us consider @xmath6 @xmath39-probe - branes in this background , which may be described by @xmath40 .
we choose a specific configuration such that @xmath41 is constant(@xmath42 ) .
it corresponds to the maximal asymptotic separation between @xmath39 and @xmath43 . with a new variable @xmath44 , instead of @xmath45 , defined by the relation ,
@xmath46 the induced metric on d8 is @xmath47 where @xmath48 consider the @xmath49 gauge field @xmath50 on the probe d8-brane configuration .
the effective action is 9-dimensional and reads @xmath51 where @xmath52 , @xmath53 is the rr 4-form field strength and @xmath54 is the chern - simons 5-form , @xmath55 which satisfies @xmath56 .
assuming that @xmath57 and @xmath58 and @xmath59 are independent of the coordinates on the @xmath35 , the induced dbi action becomes 5-dimensional @xmath60+[f^5 ] } \ , \nn\\ \label{ff}\end{aligned}\ ] ] where @xmath61 .
@xmath62 $ ] and @xmath63 $ ] are short for structures of the type @xmath64 \sim f_{\mu z}^2 f_{\mu \nu}^2 + f_{\mu\nu}^4 \\ & & [ f^5 ] \sim f_{\mu z}^2 f_{\mu \nu}^3\end{aligned}\ ] ] the chern - simons term is @xmath65 with the normalization @xmath66 , and @xmath67 is the five - dimensional plane parameterized by @xmath68 and @xmath44 .
in this part we review how the 5-dimendional dbi action yields the 4-dimensional effective action . essentially , the 5-dimensional induced action with compact @xmath19 is 4-dimensional for all excitations with wavelengths larger than the compactification radius of the order of @xmath69 . throughout and for zero baryon density ,
our discussion parallels the original discussion by sakai and sugimoto @xcite .
we quote it for notation and completeness , and refer to their work for further details .
the leading terms in the @xmath70 expansion of the dbi action is @xmath71\ , \label{dbi}\end{aligned}\ ] ] where @xmath72 in order to extract four - dimensional meson fields out of the five dimensional gauge field , we expand the gauge field as @xmath73 we choose the functions @xmath74 to be the eigenfunctions satisfying the self - adjoint differential equation @xmath75 where @xmath76 is the eigenvalue and @xmath77 is a complete set .
the normalization is fixed by the laplacian in ( [ deqn;psi ] ) @xmath78 this implies @xmath79 the @xmath80 are chosen such that @xmath81 satisfying the normalization condition : @xmath82 which is compatible with ( [ deqn;psi ] ) and ( [ norm;psi ] ) . inserting ( [ exp1 ] ) and ( [ exp2 ] ) into ( [ dbi ] ) and using the orthonomality of @xmath83 and @xmath84 , yield in the expansion ( [ exp1 ] ) and ( [ exp2 ] )
, we have implicitly assumed that the gauge fields are zero asymptotically , i.e. @xmath86 as @xmath87 .
the residual gauge transformation that does not break this condition is obtained by a gauge function @xmath88 that asymptotes a constant @xmath89 at @xmath90 .
we interpret @xmath91 as elements of the chiral symmetry group @xmath92 in qcd with @xmath6 massless flavors . by weakly gauging the @xmath92 chiral symmetry
, we may introduce the external gauge fields @xmath93 .
for the interaction between mesons and photon @xmath94 , we may choose @xmath95 where @xmath96 is the electromagnetic coupling constant and @xmath97 is the electric matrix - valued charge @xmath98 for @xmath99 . to insert the external - source gauge fields
, we impose the asymptotic values of the gauge field @xmath58 on the d8-probe - brane as @xmath100 this is implemented by modifying the mode expansion ( [ exp1 ] ) as @xmath101 where @xmath102 are defined as @xmath103 which are the non - normalizable zero modes of ( [ deqn;psi ] ) satisfying @xmath104 .
the @xmath106 gauge can be achieved by applying the gauge transformation @xmath107 with the gauge function @xmath108 the asymptotic values of ( [ asymp ] ) change to @xmath109 and @xmath110 where @xmath111 .
the gauge field in the @xmath106 gauge can be expanded as @xmath112 the residual gauge symmetry in the @xmath106 gauge is given by the @xmath44-independent gauge transformation .
the residual gauge symmetry @xmath113 and the weakly gauged chiral symmetry @xmath114 act on these fields as @xmath115 here @xmath116 are interpreted as the @xmath49 valued fields @xmath117 which carry the pion degrees of freedom in the hidden local symmetry approach . actually the transformation property ( [ xih ] ) is the same as that for @xmath117 if we interpret @xmath113 as the hidden local symmetry .
they are related to the @xmath49 valued pion field @xmath118 in the chiral lagrangian by @xmath119 the pion field @xmath120 is identical to @xmath121 in ( [ exp2 ] ) in leading order .
is a hermitian matrix , while @xmath121 and the vector meson fields @xmath122 are anti - hermitian . ]
if we choose the gauge @xmath123 then the gauge field is @xmath124 an alternative gauge is also @xmath125 it is interesting to note with sakai and sugimoto @xcite that the induced effective action expanded to fourth order in the gradients yields exactly a skyrme - like action @xmath128 ^ 2 \right ) \ , \label{skyrme}\end{aligned}\ ] ] where the pion decay constant @xmath129 and the dimensionless parameter @xmath130 are fixed @xmath131 also in the @xmath106 gauge , the chern - simons term is generic @xmath132 where @xmath133 is the gauge function given in ( [ ginv ] ) and the 4-form @xmath134 is @xmath135 as expected ( [ wzw ] ) yields the wzw term in qcd with gauged pion fields . indeed ,
if we omit the vector meson fields @xmath136 , we have @xmath137 where @xmath138+\nn\\ & & + { { \rm tr } } [ da_rdu^{-1}a_l u-{\rm{p.c . } } ] + { { \rm tr}}[a_r(du^{-1}u)^3-{\rm{p.c.}}]+\nn\\ & & + { \frac{1}{2}}{{\rm tr}}[(a_rdu^{-1}u)^2-{\rm{p.c . } } ] + { { \rm tr}}[ua_r u^{-1}a_l dudu^{-1}-{\rm{p.c.}}]-\nn\\ & & -{{\rm tr}}[a_r du^{-1}ua_r u^{-1}a_lu-{\rm{p.c . } } ] + { \frac{1}{2}}{{\rm tr}}[(a_ru^{-1}a_lu)^2]\ . \label{z}\end{aligned}\ ] ] here `` p.c . ''
represents the terms obtained by exchanging @xmath139 . the induced dbi action ( [ dbi ] )
carries exact vector dominance . indeed , by diagonalizing the kinetic terms of the vector meson fields , vector meson dominance emerges from the underlying dbi action naturally . to see that , we recall that the gauge fields are @xmath140 and involve both the external vector sources @xmath141 and the dynamical vector fields @xmath142 . to diagonalize the kinetic terms of the dynamical vector fields require the introduction of the physical tilde vector fields @xmath143 which are @xmath144 in terms of which the gauge fields now read @xmath145 with @xmath146 the dbi action ( [ dbi ] ) in terms of the physical vector fields to fourth order is note that all the couplings between the external gauge fields @xmath148 and the vector meson fields @xmath149 vanish in the first term of the effective action ( [ dbi ] ) . the second term in the effective action ( [ dbi ] ) the mesons couple to the external gauge fields only through @xmath151 and @xmath152 transitions in ( [ lsec ] ) . finally , inserting ( [ amu ] ) and
( [ azpi ] ) into the chern - simons action gives where @xmath154 , @xmath155 it shows complete vector meson dominace in the wzw term .
the terms with two or more pion fields are absent .
for simplicity we work in euclidean space , in which case the dbi action reads @xmath156+[(f^e)^5 ] } \nn \ , \\
\label{ffe}\end{aligned}\ ] ] following prescription .
@xmath157 from now on we omit the superscript @xmath158 . from the arguments to follow
, we will see that due to vector dominance no _ matter term _ is generated through the chern - simons part . in a way this is expected ; the anomaly in the chern - simons form is an ultraviolet effect that after regularization shows up in the infrared .
since matter dwells mostly in the infrared , the scale decoupling insures the insensitivity of the chern - simons terms therefore of the anomaly .
this result is usually referred to as the non - renormalization theorem .
the density will be introduced through the boundary and via the external vector field @xmath159 through @xmath160 the master gauge field now reads from ( [ amu ] ) @xmath161 for calculational convenience below , we split @xmath162 in two parts : a constant @xmath10 part ( @xmath163 ) and the rest ( @xmath164 ) , @xmath165 where @xmath166 and @xmath164 are the constant @xmath10 part and the rest part of @xmath167 respectively .
thus @xmath168 defining , @xmath169 leads @xmath170 the corresponding field strengths are @xmath171 { { \psi_{n}}}{{\psi_{m}}}\ , \\ f_{z \mu}&\!=\ !
& i \partial_\mu\pi\,\phi_0 + \sum_{n=1}^{\infty}\{i{{x^n}}\delta_{\mu 0 } + { { \bar{b}^n_\mu}}(x^\mu)\ , \}{{\dot{\psi}_n}}+ i\sum_{n=1}^{\infty } [ \pi,{{\bar{b}^n_\mu}}]{{\psi_{n}}}{{\phi_0}}\label{fmuz } \ .\end{aligned}\ ] ] notice that @xmath172 does not contribute to @xmath173 .
the additional terms due to @xmath172 come from @xmath174 .
they contribute to the induced action through @xmath175 \ , ( { { \partial_\mu}}{{\bar{b}^l_\nu}}-{{\partial_\nu}}{{\bar{b}^l_\mu}})\right\}{{\psi_{n}}}{{\psi_{m}}}{{\psi_{l}}}\ , \label{fmunu2 } \\
f_{\mu z}f_{\mu z } & \!=\!&-({\partial}_\mu\pi{\partial}_\mu\pi ) \phi_0 ^ 2 + { { \bar{b}^n_\mu}}{{\bar{b}^m_\mu}}{{\dot{\psi}_n}}{{\dot{\psi}_m}}- { { x^n}}{{x^m}}{{\dot{\psi}_n}}{{\dot{\psi}_m}}\nn \\ & & + \left(i \ { { \partial}_\mu \pi\ , { { \bar{b}^n_\mu}}\ } - 2 ( { \partial}_0 \pi){{x^n}}\right ) \phi_0 { { \dot{\psi}_n}}+ 2 i { { x^n}}{{\bar{b}^m_0}}{{\dot{\psi}_n}}{{\dot{\psi}_m}}\nn \\ & & + \left ( - 2{{x^n}}[\pi\ , { { \bar{b}^m_0 } } ] + i\{{{\bar{b}^n_\mu}}\ , [ \pi\ , { { \bar{b}^m_\mu } } ] \ } \right ) { { \dot{\psi}_n}}{{\psi_{m}}}{{\phi_0}}+ \left\{{{\partial_\mu}}\pi \ , [ { { \bar{b}^n_\mu}}\ , \pi ] \right\}{{\phi_0}}^2 { { \psi_{n}}}\ , \label{fmunufmuz}\\ \ [ f^4 ] & \!=\ ! & ( 2\pi\alpha')^4 \frac{9}{4 } \frac{r^6}{u_{kk}u_z^2 } \left\{i({\partial}_0 \pi)\phi_0 + ( i{{x^n}}+ { { \bar{b}^n_0}}){{\dot{\psi}_n}}\right\}\frac{1}{2}f_{ij}f_{ij } \left\{i({\partial}_0 \pi)\phi_0 + ( i{{x^m}}+ { { \bar{b}^m_0}}){{\dot{\psi}_m}}\right\ } \nn \\ \ [ f^5 ] & \sim & ( 2\pi\alpha')^5{{x^n}}{{x^m}}{{\dot{\psi}_n}}{{\dot{\psi}_m}}\frac{1}{2}(f_{12}f_{23}f_{31 } - f_{31}f_{12}f_{23})\end{aligned}\ ] ] for notational simplicity we have omitted @xmath176 . the induced dbi action at finite density can be separated into field independent and dependent parts @xmath177 and @xmath178 respectively @xmath179+[(f)^5]\end{aligned}\ ] ] so that @xmath180 \nn \\ & = & s_1({{x^n } } ) + s_2[\pi , v_\mu ; { { x^n } } ] + \cdots \ , \nn\end{aligned}\ ] ] with @xmath181 & \equiv & { \widetilde}t \int d^4x dz\ , u_z^2 { { \rm tr}}\ , \left[{\frac{1}{2}}\frac{p_1}{\sqrt{p_0 } } -\frac{1}{8 } \frac{p_1 ^ 2}{\sqrt{p_0}^3 } + \cdots \right]\ .\end{aligned}\ ] ] we note that the dependence on the chemical potential @xmath182 in @xmath183 is of the the type @xmath184 .
there is no term linear in @xmath182(except through the chern - simons term which is odd under @xmath185 ) .
the reason is that the qcd partition function is even under @xmath186 since the matter spectrum is symmetric around zero quark virtuality .
we believe that this behavior maybe derived geometrically from a change in the underlying metric , but we are unable to show it . the chemical potential @xmath182 as defined in ( [ amu ] ) with ( [ boundary])(i.e .
@xmath187 ) yields @xmath188 .
thus there is no contribution to the dbi action .
the only contribution stems from the chern - simons action as can be checked explicitly
. however the definition ( [ amu ] ) is at odd with vmd as explained in detail in @xcite . in other words
the physical vector field are @xmath189 defined in ( [ wtv ] ) and not @xmath190 in ( [ amu ] ) .
so ( [ amu ] ) has to be substituted by ( [ extexp33 ] ) . in this case @xmath191 and @xmath182 contributes to the dbi action but not to the chern - simons action .
on the boundary the induced dbi action describes a dense hadronic system at finite baryon density whereby the baryons are skyrmions in large @xmath1 .
the free energy @xmath192 is given as @xmath193{{\cal d}}[\pi ] e^{-s_{dbi}^e}\right ) \nn \\ & = & -kt \mathrm { ln } \left ( n ' \int { { \cal d}}[v_\mu]{{\cal d}}[\pi ] e^{-(s_1({{x^1 } } ) + s_2[\pi , v_\mu;{{x^n}}])}\right ) \nn \\ & = & ts_1 -t \mathrm { ln } \left(\int { { \cal d}}[v_\mu]{{\cal d}}[\pi ] e^{-s_2[\pi , v_\mu;{{x^n } } ] ) } \right)\end{aligned}\ ] ] where @xmath194 is the grand partition function and @xmath195 is an irrelevant matter independent constant .
the gibbs relation @xmath196 implies that @xmath197 is the free energy at zero temperature and finite baryon density . ignoring the meson zero - point contributions ( classical limit ) ,
the leading contribution to the free energy per unit volume ( minus the pressure ) is @xmath198 where @xmath199 , @xmath200 , and @xmath201 .
+ in terms of the meson physical parameters fixed at zero density , in the physical quantity , like @xmath202 , means the value at zero density . ]
@xmath203 the free energy density becomes @xmath204 with @xmath205 the sum over the vector meson - species @xmath206 is subsumed .
( [ ee ] ) receives corrections from the quantum loops with mesonic insertions . in a way ( [ ee ] )
captures the deformation of the underlying d8-probe - brane in the presence of a fixed external @xmath14 to leading order .
the square root shows that the response is not analytic , with a subtle branch point singularity to be physically interpreted below . in the limit @xmath207 but fixed baryon density @xmath208 ( low density limit ) , @xmath209
the free energy density can be tayor expanded as @xmath210 which is quadratic in @xmath211 and of order @xmath212 which is small . in the limit @xmath207 but fixed quark density @xmath213 ( high density limit ) , @xmath214 so that @xmath215 which is of order @xmath216 . in fig .
[ grandpotx ] we show the free energy versus @xmath217 for @xmath218 and @xmath219 , @xmath220 mev . since @xcite @xmath221 only the @xmath222 contribution is justified at low energy .
so @xmath223 .
here @xmath224 plays the role of a gauge - shifted chemical potential . ,
width=453 ] the baryon density @xmath225 follows from the free energy through @xmath226 where again only the @xmath222 contribution was retained .
this is a transcendental equation that reexpresses the shifted chemical potential @xmath224 in terms of the physical baryon density @xmath225 . for large @xmath1 and fixed @xmath211
it shows that @xmath227 which we interpret as the low density phase . for large @xmath1 and fixed @xmath228
it shows that @xmath229 which we interpret as the high density phase . in fig .
[ nbratio ] we show the numerical solution to ( [ integral ] ) for @xmath230 versus @xmath231 , where @xmath232 is nuclear matter density .
asymptotically large densities are attained for a critical @xmath233 . for @xmath234 and @xmath235
we have @xmath236 and @xmath237 .
the existence of a limiting chemical potential reflects on the fact that the underlying d8-probe - brane becomes unstable if the external @xmath14 becomes very large , as the square root in the ground state energy developes an imaginary part @xmath238 .
vs @xmath239,width=453 ] the pressure is @xmath240 which is @xmath241 in fig . [ press ]
we show how the pressure changes with baryonic density @xmath230 for @xmath234 , @xmath235 , @xmath242 mev and @xmath220 mev .
the pressure is dominated by _ repulsive two - body _ skyrmion - skyrmion like interactions at low density(the term of order @xmath243 in ( [ eq : pressure ] ) ) . at high density
_ attractive three - body _ skyrmion - skyrmion - skyrmion like interactions appear and cause the pressure to saturate(the term of order @xmath244 in ( [ eq : pressure ] ) ) .
the two - body interaction are 100 times stronger throughout the density range explored .
the specific fit is @xmath245 if we recall that in matter the kinetic energy @xmath246 , then the slope in fig .
[ press ] is a measure of the kinetic energy .
the holographic principle suggests that hadronic matter at large @xmath1 freezes at @xmath247 .
, width=453 ] the energy density is defined as @xmath248 and is sensitive to the gauge choice of the _ dynamical _ @xmath0 vector meson ( omega ) . in the presence of @xmath11 mesons , @xmath189
, the energy density is dependent on the constant part of the @xmath11 meson field , i.e. @xmath249 .
this is reminicent of the gauge sensitivity of the em potential . in the gauge
where @xmath250 , the energy density is shown in fig .
[ eenergy ] .
the repulsive three - body interactions soften the rise in the energy density .
note that the rise becomes almost linear after @xmath251 .
the slope is sensitive to the choice of the gauge @xmath250 .
since the pressure curve shows total freezing at these large densities , we expect that hadronic matter crystallizes .
the slope is expected to be _ negative _ and a measure of _ madelung constant _ for the specific crystal symmetry .
, width=453 ]
the dbi effective action in matter fixes completly the meson dispersion laws and interactions . indeed , in matter @xmath252 $ ] is @xmath253 & = & \int d^4x dz\ , { { \rm tr}}\ , \left [ q { \widetilde}t u_{kk } u_z^2 { p_1 } - \frac{1}{4 } q^3 { \widetilde}t u_{kk } u_z^2 p_1 ^ 2 + \cdots \right]\ .\end{aligned}\ ] ] where @xmath254 @xmath255 only the term @xmath222 in the summation over the vector meson species was retained ( see ( [ eigenvalues ] ) ) . in this section
we treat @xmath58 as anti - hermitiaon matrices with generators @xmath256 normalized as @xmath257 .
the effective action is rotated back to minkowski space . throughout
, only terms up to the third order in the fields are retained , i.e.@xmath258 .
the leading terms to this order in @xmath63 $ ] vanishes because of the cyclicity of the trace . the expansion can be simplified by using the notations @xmath259 so that the last two contributions in ( [ pirho1 ] ) vanish because @xmath261 , @xmath97 , @xmath262 and @xmath263 are even functions and @xmath264 is an odd function .
the new two terms related to @xmath265 are in the 6th and 7th lines of ( [ pirho1 ] ) .
all the parameters appearing in ( [ pirho1 ] ) are tabulated below for non - zero density .
their respective values at zero density are recorded on the right - most column for comparison .
we note that @xmath266 for fixed @xmath182 , so that all coefficients are of order @xmath267 for fixed @xmath182 at large @xmath1 . [
cols="^,<,^",options="header " , ] for @xmath234 , @xmath268mev , @xmath269mev , we show the pion in fig .
[ fpi2tsnormpi ] and vector meson in fig .
[ fpi2tsnormrho ] renormalization constants , both for the time and space - components .
the resulting velocity of the pion and the rho meson in holographic dense matter is shown in fig .
[ fpi2tsvelo ] . the pion velocity @xmath270 with @xmath271 , the spatial ( @xmath272 ) and temporal ( @xmath273 )
pion decay constants , approaches zero when hadronic matter freezes .
the vector meson velocity @xmath274 is about half . and
@xmath275,title="fig:",width=264 ] and @xmath275,title="fig:",width=264 ] and @xmath276 , title="fig:",width=264 ] and @xmath276 , title="fig:",width=264 ] ( right),title="fig:",width=264 ] ( right),title="fig:",width=264 ] the pion is massless throughout as the current quark masses are set to zero in this work .
the vector meson masses follow by higgsing , and change in matter both space - like and time - like .
space - like , the vector meson masses are screening masses .
the unrenormalized masses increase with increasing baryon density as shown in fig .
[ fpi2tsunr ] .
the physical screening masses(@xmath277 ) are obtained by normalizing with the pertinent matter dependent _ space - like _ wavefunction renormalizations .
both the vector and isovector screening masses increase with increasing baryon density as they should .
[ fpi2tsscreening ] show the screening masses for longitudinal ( left ) and transverse ( right ) vectors .
they are defined as @xmath278 and @xmath279 , title="fig:",width=264 ] and @xmath279 , title="fig:",width=264 ] the time - like vector masses are obtained similarly to the screening masses by instead using the _ time - like _ wavefunction renormalizations .
that is @xmath280 the dependence of @xmath281 on the baryon density is shown in fig .
[ fpi2tsmass ] .
the rho meson mass drops by 20% when holographic dense matter freezes .
the @xmath284 coupling is modified in matter .
the longitudinal and transverse couplings are shown in fig .
[ fpi2tsvpp ] ( left ) .
the longitudinal coupling drops by 40% at the freezing point , while the transverse coupling _ vanishes _ when dense matter freezes . in fig .
[ fpi2tsvpp ] ( right ) @xmath283 coupling also decrease in matter .
the transverse couplings drop by 90% , while the longitudinal ones by 30% .
in addition to these standard couplings , new matter dependent couplings emerge .
their composition and dependence on the baryon density is shown in fig .
[ fpi2tsnew ] .
many of the results presented here in matter bears similarities ( and differences ) with arguments presented at finite temperature using phenomenological models with hidden local symmetry @xcite . and @xmath283 couplings.,title="fig:",width=264 ] and @xmath283 couplings.,title="fig:",width=264 ]
we have provided a minimal extension of holographic qcd @xcite to dense matter .
the induced dbi effective action on d8 in the presence of a constant external @xmath0 source @xmath14 is the effective action we expect from the gravity dual theory from general principles . indeed ,
if we were to solve for the effect of the boundary @xmath14 on the d4-brane background and the resulting dbi action on the d8-probe brane , the answer is the _ externally gauged _ dbi induced action in the vacuum .
the effective action develops an imaginary time at large @xmath14 , a signal that the d8-probe - brane craks in the external field .
holographic matter describes dense qcd at large @xmath1 with baryons as solitons .
its bulk pressure asymptotes a constant at large density , signalling total freezing with zero mean kinetic energy . before freezing the matter
is dominated by two - body repulsion at low density and three - body attraction at intermediate densities .
the two - body effects are 100 times stronger than the three - body effects . in holographic matter the pions stall to almost a stop , while the vector mesons only slow down .
the vector masses drop by about 20% , while vector screening becomes increasingly large .
the transverse vector mesons completly decouple from pions at large densities .
many of the current results bear similarities with known results from effective models at large @xmath1 . in a way
, they are new as they provide first principle calculations to large @xmath1 and strongly coupled qcd . in the work to follow
, we will present results for holographic matter including finite quark masses and temperature .
the work of kyk and iz was supported in part by us - doe grants de - fg02 - 88er40388 and de - fg03 - 97er4014 .
the work of sjs was supported by kosef grant r01 - 2004 - 000 - 10520 - 0 and by src program of the kosef with grant number r11 - 2005- 021 . to mention some of them : + j. polchinski and m. j. strassler , phys .
* 88 * ( 2002 ) 031601 ; hep - th/0109174 ; + a. karch and e. katz , jhep * 0206 * , 043 ( 2002 ) [ arxiv : hep - th/0205236 ] ; + j. babington , j. erdmenger , n. j. evans , z. guralnik and i. kirsch , phys .
d * 69 * , 066007 ( 2004 ) [ arxiv : hep - th/0306018 ] .
+ m. kruczenski , d. mateos , r. c. myers and d. j. winters , jhep * 0405 * , 041 ( 2004 ) [ arxiv : hep - th/0311270 ] . t. sakai and s. sugimoto , prog .
phys . * 113 * , 843 ( 2005 ) [ arxiv : hep - th/0412141 ] ; + t. sakai and s. sugimoto , prog .
* 114 * , 1083 ( 2006 ) [ arxiv : hep - th/0507073 ] .
j. m. maldacena , _ adv .
* 2 * , 231 ( 1998 ) _ , _ int .
j. theor . phys .
* 38 * , 1113 ( 1999 ) _ [ arxiv : hep - th/9711200 ] , for review , see : + o. aharony , s. s. gubser , j. m. maldacena , h. ooguri and y. oz , phys . rept . *
323 * , 183 ( 2000 ) [ arxiv : hep - th/9905111 ] .
g. policastro , d. t. son and a. o. starinets , phys .
* 87 * ( 2001 ) 081601 ; + s. j. sin and i. zahed , phys . lett .
* b608 * ( 2005 ) 265 ; + e. shuryak , s. j. sin and i. zahed , arxiv : hep - th/0511199 ; + c.p .
herzog , a. karch , p. kotvun , c. kozcaz , and l. yaffe , hep - th/0605158 ; + s. gubser , hep - th/0605182 . to list some of them ; + a. chamblin , r. emparan , c. v. johnson and r. c. myers , phys .
d * 60 * , 064018 ( 1999 ) [ arxiv : hep - th/9902170 ] ; + k. maeda , m. natsuume and t. okamura , phys .
d * 73 * , 066013 ( 2006 ) [ arxiv : hep - th/0602010 ] ; + d. t. son and a. o. starinets , jhep * 0603 * , 052 ( 2006 ) [ arxiv : hep - th/0601157 ] ; + s. s. gubser and j. j. heckman , jhep * 0411 * , 052 ( 2004 ) [ arxiv : hep - th/0411001 ] . | we provide a method to study hadronic matter at finite density in the context of the sakai - sugimoto model .
we introduce the baryon chemical potential through the external @xmath0 in the induced ( dbi plus cs ) action on the d8-probe - brane , where baryons are skyrmions .
vector dominance is manifest at finite density .
we derive the baryon density effect on the energy density , the dispersion relations of pion and vector mesons at large @xmath1 .
the energy density asymptotes to a constant at large density suggesting that dense matter at large @xmath1 freezes , with the pion velocity dropping to zero .
holographic dense matter enforces exactly the tenets of vector dominance , and screens efficiently vector mesons . at the freezing point
the @xmath2 coupling vanishes with a finite rho mass of about 20% its vacuum value .
hep - th/0608046 * dense hadronic matter in holographic qcd * + keun - young kim@xmath3 , sang - jin sin @xmath4 and ismail zahed@xmath3 + @xmath3 _ department of physics and astronomy , suny stony - brook , ny 11794 _
@xmath4 _ department of physics , hanyang university , seoul 133 - 791 , korea _ + |
ideal hydrodynamic models have been successful @xcite for the quark - gluon plasma ( qgp ) created at the relativistic heavy ion collider ( rhic ) in brookhaven national laboratory ( bnl ) .
our next step is to develop viscous hydrodynamic models because the success suggests the deviation from the equilibrium is not so large .
we mainly consider the effects of bulk viscosity since the importance of bulk viscosity in terms of the violation of scale invariance near the crossover temperature @xmath0 has been re - recognized @xcite .
particle spectra are affected by viscosity in two ways : one is the variation of the flow @xmath1 and the hypersurface @xmath2 , the other is the distortion of the distribution @xmath3 because one needs to employ the cooper - frye formula @xmath4 at freezeout to convert the hydrodynamic results into the particle picture . here
the index @xmath5 denotes particle species and @xmath6 the degeneracy .
this enables us to follow further dynamical evolution of the hot qcd matter in the cascade models . in this study
we concentrate on corrections to particle spectra due to the modification of the distribution at freezeout in the framework of grad s method .
we express the distortion of the distribution @xmath7 in terms of macroscopic variables for a relativistic multi - component gas within the framework of the grad s 14-moment method . for this
we generalize the method developed in ref .
@xcite for a single component system by specifying non - trivialities proper to a system of multi - species particles .
the details are shown in ref .
@xcite .
we have 14 equations that relate the macroscopic variables with the microscopic distribution @xmath8 : the kinetic definitions of the bulk pressure @xmath9 , the energy current @xmath10 , the charge current @xmath11 and the shear stress tensor @xmath12 \delta t^{\alpha \beta}$ ] and the two landau matching conditions @xmath13 and @xmath14 . here
@xmath15 and @xmath16 are the dissipative parts of the energy - momentum tensor and the baryon number flux .
@xmath17 is the flow and @xmath18 is the projection operator where the metric @xmath19 .
we need the landau matching conditions to ensure the thermodynamic stability to the first order .
this can be shown by explicitly calculating @xmath20 , where @xmath21 is the entropy current .
these equations can be written in terms of @xmath7 by using the expressions of @xmath15 and @xmath16 in the relativistic kinetic theory for a multi - component system , @xmath22 where @xmath23 is the baryon number .
the grad s 14-moment method allows us to write the distortion of the distribution for the @xmath5-th particle species with 14 unknowns as @xmath24 where the positive sign is for bosons and the negative one for fermions . in the conventional formalism
one has the scalar term @xmath25 , but here we do not have it because the traceless condition is not imposed on the tensor term @xmath26 .
these two formalisms are not equivalent for a multi - component system because the scalar term in the conventional method has no particle species dependence whereas the trace part in our method does in the sense that it is dependent on the mass of the @xmath5-th particle species .
we obtain three independent sets of equations by inserting @xmath7 into the conditions : @xmath27 where @xmath28 is a coefficient of the expansion of the moment @xmath29.\end{aligned}\ ] ] @xmath30 is defined in the same way but with the baryon number @xmath23 as an additional weight factor in the summation over the particle species index @xmath5 .
the unique form of @xmath7 can be obtained by solving them .
the correction tensors are @xmath31 where the prefactors @xmath32 and @xmath33 are functions of @xmath28 and @xmath30 .
one can not take the scalar term in ( [ eq : df ] ) because that induces a change of the sign in the bulk pressure - related prefactors @xmath34 , @xmath35 and @xmath36 which causes divergence in the case of the 16-component hadron resonance gas discussed later .
note here that ( i ) information of all the components in the gas comes into the viscous modification of a particle species through the transport coefficients , @xmath28 s and @xmath30 s , and ( ii ) if one takes the quadratic ansatz in ref .
@xcite , _ i.e. _ , @xmath37 , the system becomes thermodynamically unstable because the matching conditions are not satisfied .
-spectra and ( b ) elliptic flow coefficients @xmath38 for the negative pion .
solid , dotted , and dashed lines are , respectively , the results without any viscous corrections , with the effect of bulk viscosity with @xmath39 and with @xmath40 . the thick(thin ) lines show that the absolute value of the ratio of the correction to the ideal spectrum is smaller(larger ) than 0.5 . , title="fig:",scaledwidth=49.0% ] -spectra and ( b ) elliptic flow coefficients @xmath38 for the negative pion .
solid , dotted , and dashed lines are , respectively , the results without any viscous corrections , with the effect of bulk viscosity with @xmath39 and with @xmath40 . the thick(thin ) lines show that the absolute value of the ratio of the correction to the ideal spectrum is smaller(larger ) than 0.5 .
, title="fig:",scaledwidth=49.0% ] one needs models for the equation of state ( eos ) , the macroscopic variables and profiles of the flow and the hypersurface to estimate the effects of @xmath41 on particle spectra .
we consider 16-component hadron resonance gas for the eos , which has mesons and baryons with the mass up to @xmath42 .
the bulk pressure is estimated with the navier - stokes limit , _
i.e. _ , @xmath43 where the models of bulk and shear viscosity are @xmath44 @xcite and @xmath45 @xcite . here
@xmath46 and @xmath47 are the entropy density and the sound velocity , respectively .
@xmath48 is a free parameter .
the freezeout temperature @xmath49 is set to @xmath50 gev where @xmath51 and @xmath52 when @xmath40 .
the flow and the hypersurface are taken from a ( 3 + 1)-dimensional ideal hydrodynamic simulation @xcite in au+au collisions at @xmath53 gev where the impact parameter is 7.2 fm .
it should be noted again that what we estimate here is not the distortion of the flow @xmath54 but the viscous corrections through the modification of the distribution @xmath41 at freezeout . for numerical estimations
we employ the zero net baryon density limit , which is well justified for the hot qcd matter at rhic .
we do not have to worry about the loss of the landau matching condition of the baryon number flux because the condition still yields a finite relation even in the limit of vanishing chemical potential @xcite .
we estimate the particle spectra of negative pion with mass @xmath55 gev using the cooper - frye formula ( [ eq : ptspectra ] ) without changes of flow velocity nor freezeout hypersurface from the ideal hydrodynamic results .
the bulk viscosity lowers the mean @xmath56 of the particle spectra , @xmath57 , when estimated with the ideal hydrodynamic flow and hypersurface .
this can be attributed to the reduction of the effective pressure due to the bulk viscosity .
the isotropic nature of the bulk pressure reduces the pressure in transverse directions when the system is expanding in longitudinal direction . on the other hand
, the bulk viscous correction enhances the elliptic flow coefficients @xmath38 .
this can be interpreted in the relation @xmath58 @xcite , because the slope of differential @xmath38 is expected to become steep when the @xmath57 is decreased while the average @xmath59 is not affected by viscosity with in our calculations .
we have to be aware that the viscous corrections might have been overestimated because ( i ) we take the navier - stokes limit for @xmath60 and thus do not consider relaxation effects , and ( ii ) the derivatives of ideal hydrodynamic flow is generally larger than those of real viscous hydrodynamic flow as viscosity tends to eliminate differences in the flow .
we obtained the unique and consistent expression of the distortion of the distribution for a multi - component system in the framework of the grad s 14-moment method .
we find several non - trivial issues along the course of discussion .
first , one needs to introduce the non - zero trace tensor term in the correction instead of the conventional scalar term in the case of the 16-component hadron resonance gas to get a finite result .
second , the distortion is different by particle species , but is dependent on all the components in the gas because the transport coefficients and macroscopic quantities @xmath28 and @xmath30 are .
third , the matching for the baryon number flux remains meaningful even in the limit of vanishing chemical potential .
effects of bulk viscosity on particle spectra through the distortion of the distribution is estimated , and we find it suppresses @xmath56-spectra and enhances @xmath38 when estimated with ( 3 + 1)-dimensional ideal hydrodynamic flow and hypersurface
. we should be careful of the possible overestimation in the usage of the navier - stokes limit and the ideal hydrodynamic flow .
a ( 3 + 1)-dimensional viscous hydrodynamic model is necessary to accurately estimate the viscous effects to compare with experimental data and to constrain the transport coefficients from them .
the authors acknowledge fruitful discussions with t. hatsuda , t. kodama , t. kunihiro , and s. muroya .
the authors are also grateful for valuable comments by g. s. denicol .
the work of t.h . was partly supported by grant - in - aid for scientific research no .
19740130 and by sumitomo foundation
080734 . 00
t. hirano , phys .
c 65 ( 2002 ) 011901 .
t. hirano , k. tsuda , phys . rev .
c 66 ( 2002 ) 054905 .
d. kharzeev , k. tuchin , jhep 0809 ( 2008 ) 093 .
w. israel , j. m. stewart , annals phys .
118 ( 1979 ) 341 .
a. monnai , t. hirano , arxiv:0903.4436 [ nucl - th ] .
k. dusling , d. teaney , phys .
c 77 ( 2008 ) 034905
. s. weinberg , astrophys .
j. 168 ( 1971 ) 175 .
a. hosoya , m. a. sakagami , m. takao , annals phys . 154
( 1984 ) 229 .
p. arnold , c. dogan , g. d. moore , phys . rev .
d 74 ( 2006 ) 085021 .
p. kovtun , d. t. son , a. o. starinets , phys .
rev . lett .
94 ( 2005 ) 111601 .
t. hirano , m. gyulassy , nucl .
phys . a 769 ( 2006 ) 71 . | we estimate the effects of viscosity on the phase space distribution appearing in the cooper - frye formula within the framework of the grad s fourteen moment method and find that there are non - trivialities in the discussion of a multi - component system .
we calculate the viscous corrections of particle spectra and elliptic flow coefficients from the distortion of the distribution using the flow and the hypersurface taken from a ( 3 + 1)-dimensional ideal hydrodynamic simulation .
we see that the bulk viscosity have visible effects on particle spectra .
the results suggest we should treat _ both _ shear and bulk viscosity carefully when constraining the transport coefficients from experimental data . |
one of the main guiding principles in seeking an appropriate theory of formation of grains in metals and alloys during and after cold working processes producing severe plastic deformations has first been proposed by hansen and kuhlmann - wilsdorf @xcite in form of the so - called leds - hypothesis : the dislocation structures in the final state of deformation minimize the energy of crystals ( see also @xcite ) .
the main reason why the formation of grains becomes energetically preferable at severe plastic deformations lies in the non - convexity of the energy of crystal in this range @xcite . within the conventional crystal plasticity considered in @xcite
the minimization of such non - convex energy leads immediately to the infinitely fine lamellar structure with grain boundaries as sharp interfaces .
however , as mentioned by kuhlmann - wilsdorf and hansen @xcite , typical grain boundaries , termed geometrically necessary boundaries , have as a rule a non - vanishing thickness and may contain a large number of dislocations and thus contradict the conventional crystal plasticity .
the question then arrises in this connection : what kind of continuum model may resolve this conflict ?
the present paper proposes a dislocation model of formation of grains within the continuum dislocation theory @xcite which predicts the existence of such geometrically necessary boundaries . by including the energy of dislocation network containing the gradient of the plastic slip into the energy functional we regularize the non - convex energy minimization problem .
we illustrate the application of the theory on the example of single crystal deforming in single slip under a simple shear .
we show that the geometrically necessary boundaries , in which the transition from one grain to the next occurs smoothly , have a small but finite thickness and contain a large number of dislocations .
although the resultant burgers vector of dislocations in such grain boundary is non - zero , they do not produce long range stresses , and the lamellar structure of grains is in fact the low energy dislocations structure .
we also compute the thickness of geometrically necessary boundaries and their energies as functions of the misorientation angles .
based on these results we estimate also the number of grains in terms of the specimen sizes .
we consider for simplicity an initially dislocation - free single crystal having only one active slip system . in this case
the kinematic quantities characterizing its observable deformations are the placement field @xmath0 and the plastic slip field @xmath1 .
the incompatible plastic deformation is given by @xmath2 with the pair of constant and mutually orthogonal unit lattice vectors @xmath3 and @xmath4 denoting the slip direction and the normal to the slip planes .
using the multiplicative resolution of the total compatible deformation gradient @xmath5 into the plastic and elastic part @xcite , we find the incompatible elastic deformation in the form @xmath6 the tensor of dislocation density measuring the incompatibility of @xmath7 reads ( see @xcite ) @xmath8 if , in addition , all dislocation lines are straight lines parallel to the unit vector @xmath9 , then the scalar dislocation density ( or the number of excess dislocations per unit area perpendicular to @xmath9 ) can be determined as @xmath10 with @xmath11 being the magnitude of burgers vector . for crystals having as a rule small elastic strains we propose the free energy per unit volume of the undeformed configuration in the form
@xmath12 here @xmath13 corresponds to the elastic strain tensor , @xmath14 and @xmath15 are the lam constants , @xmath16 a material constant , and @xmath17 can be interpreted as the saturated dislocation density .
the first two terms in represent energy of crystal due to the macroscopic elastic deformation .
the last term describes energy of the dislocation network for moderate dislocation densities .
note that for small or extremely large dislocation densities close to the saturated density the logarithmic energy proposed in @xcite is more appropriate .
we deform this crystal occupying in the initial configuration some region @xmath18 of three - dimensional space by placing it in a displacement - controlled device such that , at the boundary @xmath19 , the conditions @xmath20 are specified , with @xmath21 being a given overall deformation . if the deformation process is isothermal , no body force acts on this crystal , and the resistance to the dislocation motion can be neglected , then the following variational principle turns out to be valid : the true placement vector @xmath22 and the true plastic slip @xmath23 in the _ final _ equilibrium state of deformation minimize the energy functional @xmath24=\int_{\mathcal{v}}w(\mathbf{f},\beta , \nabla \beta ) \ , dx\ ] ] among all _ continuously differentiable _
fields @xmath0 and @xmath1 satisfying constraints , where @xmath25 .
we will see that , due to the non - convexity of the free energy density and the presence of @xmath26 in the energy functional via the energy of the dislocation network , the formation of grains with regular grain boundaries having a finite thickness is energetically preferable .
consider now the special case of plane strain simple shear of the specimen in form of a cuboid of height @xmath27 , width @xmath28 , and depth @xmath29 such that @xmath30 , while @xmath31 , @xmath32 and @xmath1 depend only on two cartesian coordinates @xmath33 and @xmath34 and satisfy at the side boundary the conditions @xmath35 with @xmath36 being the overall shear strain .
we assume that @xmath37 , @xmath38 and all dislocation lines are parallel to the @xmath39-axis , so @xmath40 . if the deformations are uniform such that @xmath41 with @xmath42 and @xmath43 being the constants , then the energy normalized by @xmath44 and minimized with respect to @xmath43 turns out to be non - convex for @xmath45 as shown in fig .
[ fig : condensed ] ( see @xcite ) . and
the dimensionless shear stress @xmath46 for @xmath47.,width=340 ] in view of this non - convexity , we proposed in @xcite the energy minimizing sequence consisting of layers having the uniform states a and b according to @xmath48 such that the volume fraction of the layer @xmath49 is given by @xmath50 .
it has been shown in @xcite that such candidates for the minimizer satisfy the equations of equilibrium in each layer as well as the outer boundary conditions except at the side boundaries @xmath51 of the specimen . besides , the energy of such lamellar structure is equal to zero which is the minimal possible value . however
, if the boundaries between layers are sharp interfaces , these candidates do not belong to the set of admissible fields of our original variational problem and due to the jumps of @xmath52 and @xmath43 on those interfaces , so they fail to be the energy minimizers of and . to correct the behavior of those candidates for minimizers we assume now that the layers corresponding to the states a and b are separated by a thin layer of small thickness @xmath53 in which the placement and plastic slip change smoothly from state a to state b ( see fig . [ fig : grainboundarylayer ] ) .
since this boundary layer is thin ( @xmath53 is much smaller than @xmath54 , @xmath55 , and the sizes of the specimen ) , it is reasonable to assume that the displacement in @xmath34-direction is zero , while the displacement in @xmath33-direction and the plastic slip depend only on @xmath34 : @xmath56 with this ansatz it is easy to show that the determination of functions @xmath57 and @xmath58 as well as the unknown boundary layer reduces to minimizing the following functional @xmath59 = \int_{l_1}^{l_2}\left[f\left(u_{,2},\beta\right ) + g\left(\beta_{,2}\right)\right ] dx_2\ ] ] among functions @xmath57 ,
@xmath58 and unknown lengths @xmath60 , @xmath61 such that @xmath62 although the analytical solution can be found in the most general case with @xmath63 , we choose @xmath64 for the shortness of presentation . in this special case @xmath65 and @xmath66
are given by @xmath67 ^ 2\right .
\notag \\ & \left .
+ \left[\left(\beta u_{,2 } \sin\varphi \cos\varphi + \beta \cos^2 \varphi - u_{,2}\right)^2 + \left(\beta \sin\varphi \cos\varphi - 1\right)^2 - 1\right]^2\right . \\ & \left .
+ 2 \left[\left(-\beta u_{,2 } \sin\varphi \cos\varphi - \beta \cos^2 \varphi + u_{,2}\right ) \left(\beta u_{,2 } \sin^2\varphi + \beta \sin\varphi \cos\varphi + 1\right ) \right.\right . \notag \\ & \left.\left .
+ \beta \sin^2 \varphi \left(1 - \beta \sin\varphi \cos\varphi\right)\right]^2\right\ } , \quad g\left(\beta_{,2}\right ) = \frac{1}{2 } k \frac{\beta_{,2}^2\sin ^2\varphi } { b^2\rho_s ^2}. \notag\end{aligned}\ ] ] functional can be reduced to the functional depending only on @xmath43 .
indeed , the variation of with respect to @xmath57 leads to the equation and boundary conditions that imply @xmath68 so the stresses inside the boundary layer are zero .
since it was already established in @xcite that the stresses in layers a and b are also zero , the whole specimen is stress - free . solving the above equation with respect to @xmath69
we express it in terms of @xmath43 @xmath70 substituting this back into the functional we reduce the latter to the functional depending only on @xmath58 and @xmath60 , @xmath61 @xmath71 = \int_{l_1}^{l_2}\left[p\left(\beta\right ) + g\left(\beta_{,2}\right)\right ] dx_2,\ ] ] where @xmath72 ^ 2}\right.\\ & \left .
+ \frac{\left(\beta^2 - 4\right ) \beta^2 \cos 4\varphi + 12\beta^2 - 16\beta \sin 2\varphi + 16}{8\left[\beta^2 \left(1 - \cos 2\varphi\right ) + \beta^2 - 2
\beta \sin 2\varphi + 2\right]^2}\right\ } .\end{aligned}\ ] ] varying functional with respect to @xmath58 , @xmath60 , @xmath61 and taking into account the constraints we obtain the euler equation @xmath73 which is subjected to the boundary conditions @xmath74 the last two boundary conditions mean that the dislocation density must vanish at the boundaries @xmath75 and @xmath76 .
equation possesses the first integral @xmath77 using the boundary conditions as well as the identities @xmath78 , we find that @xmath79 .
the phase portrait computed for @xmath80 m , @xmath81m@xmath82 , @xmath83 and @xmath84 is shown in fig .
[ fig : phaseportrait ] .
we see that the phase curve connecting the states a and b and satisfying the boundary conditions is the separatrix in the lower half of the phase plane @xmath85 represented by the red curve with an arrow denoting the direction of change of @xmath43 as @xmath34 changes from @xmath60 to @xmath61 .
it is also easy to show that the phase curve connecting b and a is the separatrix in the upper half of the phase plane , so the dislocations in the boundary layer between b and a have another sign than those connecting a and b. versus the dimensionless coordinate @xmath86.,width=340 ] the first integral with @xmath79 enables one to find @xmath43 implicitly in terms of @xmath34 according to @xmath87 the plot of @xmath43 versus the dimensionless coordinate @xmath88 for the above chosen parameters is shown in fig .
[ fig : beta ] .
it is seen that , for large @xmath89 , the plastic slip remains in the very close neighborhood of the state a with @xmath90 or the state b with @xmath91 , while the strong change of @xmath43 leading to the transition from a to b occurs in a finite interval of @xmath92 .
( in nanometer ) versus the misorientation angle @xmath93 ( in degrees).,width=302 ] it turns out that if we compute the thickness of the boundary layer according to the formula @xmath94 then this thickness becomes infinite , so the result contradicts our assumption about the smallness of @xmath53 as compared with @xmath54 , @xmath55 , and with the sizes of the specimen .
the resolution of this conflict should be found in the discreteness of crystals that may accommodate only a large but finite natural number of dislocations .
since the dislocation density is @xmath95 , the total number of dislocations in the boundary layer equals @xmath96 as the smallest number of dislocation is 1 , the smallest quantum of plastic slip leading to the recognizable change of @xmath97 must be @xmath98 .
now , if we take @xmath99 as a small positive number and compute the thickness of the boundary layer according to @xmath100 then the thickness becomes finite ( except for @xmath101 ) . since the pre - factor
, the integrand , and both limits of integration depend on the orientation of slip system @xmath102 , the thickness is also a function of @xmath102 .
[ fig : thickness ] shows the dependence of the thickness of geometrically necessary boundary ( measured in nanometer ) on the misorientation angle @xmath93 ( measured in degrees , see @xcite ) for the above chosen parameters and @xmath103 ( the result turns out to be not strongly sensitive to the choice of @xmath99 ) .
it is seen that the thickness changes from 50 to 6 nanometers for misorientation angles
larger than 10@xmath104 .
for misorientation angles close to zero formula is not well - defined . in this case the dislocation model of small - angle tilt boundary considered in @xcite is more appropriate .
( in n / m ) versus the misorientation angle @xmath93 ( in degrees).,width=264 ] the minimum of the functional can easily be computed based on the analytical solution found above . using the first integral
we obtain @xmath105 since this minimum value has the meaning of the energy ( normalized by @xmath15 ) per unit area of the geometrically necessary boundary , the energy density of such boundary must be @xmath106 .
similar to , the pre - factor , the integrand , and the lower limit of integration in depend on the orientation of slip system @xmath102 , so @xmath107 must also be a function of @xmath102 . fig . [
fig : gbenergy ] shows the dependence of @xmath107 ( measured in n / m ) on the misorientation angle @xmath93 ( measured in degrees ) for the above chosen parameters and @xmath108gpa ( for aluminum ) .
in contrast to @xmath53 , the energy density of geometrically necessary boundary is well - defined for all misorientation angles . for small
misorientation angles up to 50@xmath104 the energy density of the geometrically necessary boundary changes from zero to approximately 1
n / m that agrees quite well with the value 0.625 n / m given in @xcite .
in @xcite we gave the estimation of the number of grains based on the following deliberations .
the energy of the boundary layers near the side boundaries required to satisfy the side boundary conditions turns out to be of the order @xmath109 , where @xmath110 is the thickness of one pair of layers a and b. taking into account the energy of the geometrically necessary boundaries which is of the order @xmath111 , we can estimate @xmath110 by minimizing these two contributions to the energy yielding @xmath112 this relation exhibits clearly the size effect .
we have shown in this paper that the presence of the gradient of plastic slip in the energy of the dislocation network enables one to regularize the non - convex energy minimization in the class of smooth displacements and plastic slips .
this leads to the formation of grains with geometrically necessary boundaries having a finite thickness .
let us mention the similarity of the proposed continuum theory with the static version of cahn - hilliard theory of phase separation @xcite and the recent phase - field approach @xcite .
the generalization of our model to capture the motion of geometrically necessary boundaries and the grain growth will be addressed elsewhere .
berdichevsky , continuum mech .
therm . 18 ( 2006a )
195 - 222 .
berdichevsky , scripta mater .
54 ( 2006b ) 711 - 716 .
cahn , j.e .
hilliard , j. chem .
( 1958 ) 258 - 267 . c. carstensen , k. hackl , a. mielke , p. roy .
soc . lond .
a mat . 458 ( 2002 ) 299 - 317 .
chen , ann .
32 ( 2002 ) , 113 - 140 .
n. hansen , d. kuhlmann - wilsdorf , mat .
( 1986 ) 141 - 161 .
hirth , j. lothe , theory of dislocations .
mcgraw - hill , new york , 1968 .
m. koster , m. , k.c .
nguyen , int .
j. plasticity 69 ( 2015 ) 134 - 151 .
d. kuhlmann - wilsdorf , mat .
eng . a - struc . 113
( 1989 ) 1 - 41 .
d. kuhlmann - wilsdorf , n. hansen , scripta metall .
mater . 25 ( 1991 ) 1557 - 1562 .
c. laird , p. charsley , h. mughrabi , mat .
81 ( 1986 ) 433 - 450 .
le , c. gnther , int .
j. plasticity 53 ( 2014 ) 164 - 178 .
m. ortiz , e.a .
repetto , j. mech .
solids 47 ( 1999 ) 397 - 462 .
m. ortiz , e.a .
repetto , l. stainier , j. mech .
solids 48 ( 2000 ) 2077 - 2114 .
read , w. shockley , phys .
78 ( 1950 ) 275 - 289 . | a continuum dislocation model of formation of grains whose boundaries have a non - vanishing thickness is proposed . for a single crystal deforming in simple shear the lamellar structure of grains with thin layers containing dislocations as the geometrically necessary boundaries turns out to be energetically preferable .
the thickness and the energy of this type of grain boundary are computed as functions of the misorientation angle .
dislocations , grain boundaries , crystal plasticity , single crystal , shear . |
in classical systems , statistical thermodynamics tells that macroscopic physical property ( or macroscopic structure ) in equilibrium state can be typically obtained by canonical average of @xmath0 , where @xmath1 denotes partition function , @xmath2 is temperature , @xmath3 means possible microscopic state on phase space , and @xmath4 and @xmath5 are energy and physical property ( or structure ) in state @xmath3 , respectively .
when @xmath2 goes to zero , summation is performed for the ground states having lowest energy .
when @xmath2 increases , due mainly to entropy contribution , the system can go into disordered states ( e.g. , states above the critical temperature ) . in the disordered states ,
direct determination of @xmath6 through the definition of canonical average is practically intractable , since number of possible microscopic states should astronomically increases with increase of system size .
thus , a variety of calculation techniques have been developed to effectively address @xmath6 .
one of the most successful techniques is monte carlo ( mc ) simulation with metropolis algorism,@xcite which samples important microscopic states contributing to @xmath6 , and the mc simulation have been subsequently modified such as multihistgram method , multicanonical ensembles and entropic sampling,@xcite in order to effectively sweep across the configuration space .
one of the exceptions is coherent potential approximation,@xcite where it considers the average occupation of elements with the lack of information about geometrical structure .
another is high - temperature expansion,@xcite enabling efficient estimation of energy as well as other physical properties at high temperature .
when we consider classical systems , typically their constituents are spatially constrained in various ways .
for instance , crystal lattice acts as constraint for substitutional crystalline solids , and volume and density as constraints for liquids in rigid box .
although the existing theoretical approaches amply predict macroscopic physical property and structure in equilibrium metallic and semiconductor alloys based on first - principles calculations,@xcite the role of the spatial constraint on equilibrium properties do not get sufficient attention so far .
following this fact , our recent studies@xcite find new representation of macroscopic physical properties ( including internal energy , density , and free energy ) and of macroscopic structure for disordered states , by using a few number of specially selected microscopic states .
we find that these special microscopic states depend only on the type of spatial constraint , and are independent of constituent elements , multibody interactions , and of temperature , which means that we can _ a priori _ know the structure of the special microscopic states .
we have demonstrated that this finding provides significantlly efficient and systematic prediction of equilibrium properties for multicomponent alloys based on first - principles calculation .
this important characteristics for equilibrium properties rely on the fact that statisitcal interdependence for density of microscopic states ( doms ) in configuration space appears numerically vanished at thermodynamic limit for a wide class of spatial constraints , whereas such behavior of the doms have not been quantitatively well - understood so far . in the present article , we theoretically tuckle this problem based on the random matrix with gaussian orthogonal ensemble , where corresponding statistical independence is mathematically guaranteed .
let us first briefly explain the density of microscopic states ( doms ) on configuration space for _ non - interacting _ system . as shown in fig .
[ fig : dos ] , we find in the present study that for a wide class of spatial constraints , doms for large number of constituents with multicomponent system can be universally well - characterized by a multidimensional gaussian distribution , namely , @xmath7 , \end{aligned}\ ] ] where @xmath8 denotes complete basis functions to describe possible microscopic states , @xmath9 is the @xmath10-component vector of @xmath11 where @xmath12 denotes taking arithmetic average over configuration space , and @xmath13 represents covariance matrix for @xmath14 , reflecting the spatial constraint .
this means that under a proper set of basis functions to give diagonal @xmath13 , statistical interdependence practically vanishes , although in principle , basis functions themselves are not statistically independent due to the existence of spatial constraint.@xcite b@xmath15 binary system on fcc lattice .
right : doms for bivariate coordination composed of two of the three figures in left . closed circles or dotted lines denote result from simulation uniformly sampling the configuration space , and solid curves denote analytical gaussian distributions . ] in order to address this practical disappearance of statistical interdependence , straightforward comparison of doms landspace for real systems and that for analytical gaussian distribution is practically intractable : this is because deviation in the landscapes is negligiblly small , which can be easily buried by numerical error in simulation to sweep across the vast configuration space . in the present study , to avoid this problem ,
we therefore take different strategy to employ random matrix with gaussian orthogonal ensemble , whose statistical independence is mathematically guaranteed .
the random matrix , @xmath16 , is given by @xmath17 matrix whose elements independently take normal random numbers , with its linear average and variance respectively taking 0 and 1 . in order to apply the random matrix to investigating characteristics of doms
, we read @xmath18 as number of uniformly sampling point on configuration space for real system , and @xmath19 as number of basis functions considered . with this definition , covariance matrix @xmath20 of doms constructed from the random matrix
is naturally given by @xmath21 where all diagonal element is normalized to 1 .
for real system , we employ generalized ising model@xcite on a given lattice , where any microscopic state on configuration space , @xmath22 , can be completely specified by @xmath23 in binary system where occupation of elements is specified by ising - like spin of @xmath24 , coefficient of the basis function @xmath25 can be simply given by @xmath26 where product @xmath27 is taken for `` figure '' @xmath28 along @xmath25 , consisting of corresponding lattice points , and @xmath29 denotes linear average over symmetry - equivalent figure to @xmath28 on microscopic state @xmath22 . the advantage to employ
the generalized ising model is that for binary system , we confirm that the basis functions above can provide diagonal @xmath30 for any combination of corresponding figures at equiatomic composition : we do not need to further find a set of basis functions to deagonalize @xmath30 . here , we therefore prepare equiatomic binary system on fcc lattice , where basis functions along up to 6th nearest - neighbor ( 6nn ) pairs and all triplets consisting of up to 6nn pairs are considered , resulting in 29 basis functions . in a similar fashion to the random matrix , we construct doms constructed from the 29 basis functions by performing monte carlo ( mc ) simulation to uniformly sampling microscopic states ( 500,000 mc step ) for 256 , 2048 , and 3024-atom mc - cell to see the system - size dependence of the doms .
variance of doms projected onto individual coordination is all normalized to one so that diagonal elements of resultant covariance matrix all take one .
, takes 256 , 2048 and 3200 . ]
as is already described above , since direct comparison of landscape of doms is intractable , we compare the density of eigenvalues ( doe ) of the covariance matrix obtained through random matrix and practical system . for random matrix , we first numerically construct @xmath17 random matrix where constituent elements are obtained by normal numbers as described above . here , to make comparison with the practical system ,
we respectively give @xmath31 and @xmath32 .
to check the validity , we compare the does of the constructed random matrix and those obtained through the characteristics of marcenko - pastur distribution , where corresponding doe of covariance matrix can be analytically determined with @xmath33 and @xmath34 , where the ratio of @xmath35 is kept fixed .
we confirm that the does for the constructed random matrix discussed below exhibits excellent agreement with the analytical ones .
figure [ fig : doe ] compares the resultant does for random matrix and for practical systems with different number of constituents , @xmath36 .
we can see that when the system size is small ( i.e. , @xmath37 ) , landscape of the doe appears completely different from that for random matrix .
especially , density of states around @xmath38 comes from the significant statistical interdependence between chosen basis functions due to the periodic boundary condition .
meanwhile , when @xmath36 increases , landscape of does for practical systems appears gradually close to that for random matrix having single sharp peak with non - zero finite width around @xmath39 .
particularly , several sub - peaks below @xmath39 for practical systems are significantly diminished when @xmath36 increases .
the highest peaks of does around @xmath39 for random matrix as well as for practical systems can be naturally attributed to the fact that trace of covariance matrix for random matrix and for practical systems take exactly the same value . to further make quantitative comparison
, we estimate lower - order ( from 2nd to 4th ) moment of the does .
figure [ fig : mom ] shows system - size dependence of the resultant 2nd , 3rd and 4th moment of does for practical system , together with that for random matrix represented by dashed horizontal lines .
we can clearly see that when @xmath36 increases , value of all moments for practical system approaches to that for random matrix .
these results therefore indicate that when the number of microscopic states around the center of gravity of doms in configuration space increase with increase of the system size , statistical interdependence along each coordination can be vanished : this certainly supports our developed approach using a few special microscopic states , which uses this characteristics of statistical interdependence to decompose contribution from each coordinate to macroscopic physical properties . to summarize , by using the characteristics of random matrix consisting of independently random numbers , we confirm that statistical interdependence of density of microscopic states in configuration space tend to disappear when the system size increases .
this result support our developed approach to provide new representation for equilibrium macroscopic properties , based on the dissapearance of the statistical interdependence for large systems .
this work was supported by a grant - in - aid for scientific research on innovative areas `` materials science on synchronized lpso structure '' ( 26109710 ) and a grant - in - aid for young scientists b ( 25820323 ) from the mext of japan , research grant from hitachi metals@xmath40materials science foundation , and advanced low carbon technology research and development program of the japan science and technology agency ( jst ) .
9 n. metropolis , a.w .
rosenbluth , m.n .
rosenbluth , a.h .
teller , and e. teller , j. chem .
* 21 * , 1087 ( 1953 ) .
a.m. ferrenberg and r. h. swendsen , phys .
lett . * 63 * , 1195 ( 1989 ) .
g. bhanot , r. salvador , s. black , p. carter , and r. toral , phys .
lett . * 59 * , 803 ( 1987 ) .
j. lee , phys .
lett . * 71 * , 211 ( 1993 ) .
m. jaros , rep .
. phys . * 48 * , 1091 ( 1985 ) .
phys . * 54 * , 235 ( 1982 ) .
n.a . zarkevich and d.d .
johnson , phys .
* 92 * , 255702 ( 2004 ) .
a. franceschetti , s.v .
dudiy , s.v .
barabash , a. zunger , j. xu and m. van schilfgaarde , phys .
lett . * 97 * , 047202 ( 2006 ) .
t. mueller and g. ceder , phys .
b * 82 * , 184107 ( 2010 ) .
j. p. julien , p.e.a .
turchi and d. mayou , phys .
b * 64 * , 195119 ( 2001 ) . a. van de walle , nature mater . *
7 * , 455 ( 2008 ) .
hart , v. blum , m.j .
walorski and a. zunger , nature mater .
* 4 * , 391 ( 2005 ) .
fischer , k.j .
tibbetts , d. morgan and g. ceder , nature mater .
* 5 * , 641 ( 2006 ) .
k. yuge , j. phys .
. jpn . * 84 * , 084801 ( 2015 ) .
k. takeuchi , r. tanaka and k. yuge , j. phys .
: condens .
matter * 27 * , 385201 ( 2015 ) .
d. de fountain , _ solid state physics _ , edited by h. ehrenreich and d. turnbull ( academic press , cambridge , massachusetts , 1994 ) , vol .
sanchez , f. ducastelle , and d. gratias , physica * 128a * , 334 ( 1984 ) . | in classical systems , our recent theoretical study provides new insight into how spatial constraint on the system connects with macroscopic properties , which lead to universal representation of equilibrium macroscopic physical property and structure in disordered states .
these important characteristics rely on the fact that statisitcal interdependence for density of microscopic states ( doms ) in configuration space appears numerically vanished at thermodynamic limit for a wide class of spatial constraints , while such behavior of the doms is not quantitatively well - understood so far .
the present study theoretically address this problem based on the random matrix with gaussian orthogonal ensemble , where corresponding statistical independence is mathematically guaranteed . using the generalized ising model
, we confirm that lower - order moment of density of eigenstates ( doe ) of covariance matrix of doms shows asymptotic behavior to those for random matrix with increase of system size .
this result supports our developed theoretical approach , where equilibrium macroscopic property in disordered states can be decomposed into individual contribtion from each generalized coordinate with the sufficiently high number of constituents in the given system , leading to representing equilibrium macroscopic properties by a few special microscopic states . |
because they are composed of old , passively evolving populations , elliptical galaxies offer great promise for tracing the evolution of the universe . whether elliptical galaxies formed through hierarchical merging or monolithic collapse , one of the major goals in extragalactic studies is the determination of the `` redshift of formation , '' @xmath5 , that marks the age where most of the stars in early - type galaxies formed .
recent studies of galaxy clusters out to @xmath6 indicate that most of the star formation had to be completed at high redshift ( @xmath7 ) , followed by quiescent evolution thereafter ( stanford , eisenhardt , & dickinson 1998 ; kodama et al .
1998 ) .
the uv upturn is a sharp rise in the spectra of e and s0 galaxies shortward of restframe 2500 .
it provides a sensitive tracer of age for the oldest stars in these galaxies , and can potentially constrain @xmath5 .
traditionally characterized by the @xmath8 color , the uv upturn in local galaxies originates in a population of hot horizontal branch ( hb ) stars and their uv - bright progeny ( see brown et al .
2000 and references therein ) .
as first noted by greggio & renzini ( 1990 ) , the uv upturn should evolve rapidly with age , through the evolution of hb morphology and the main sequence turnoff mass .
although all models of elliptical galaxy evolution predict a rapid evolution in the uv upturn ( e.g. , tantalo et al .
1996 ) , the timing for the uv upturn onset depends strongly upon model parameters .
we have been undertaking a series of observations to trace the evolution of the uv upturn as a function of redshift .
faint object camera ( foc ) observations of abell 370 ( @xmath4 ) provided the first detection of far - uv emission from quiescent elliptical galaxies above @xmath9 ( brown et al .
the very strong uv emission found at @xmath4 suggests no evolution + @xmath10based on observations with the nasa / esa hubble space telescope obtained at the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas 5 - 26555
. @xmath11 noao research associate .
+ in the uv upturn between our own epoch and one 4 gyr earlier , a finding inconsistent with some models of galaxy evolution , and apparently consistent only for high values of @xmath5 . here
, we describe observations that trace the uv upturn to higher redshift .
we used the far - uv camera on the space telescope imaging spectrograph ( stis ) to measure the uv emission from four giant elliptical galaxies in the cluster cl0016 + 16 at @xmath12 .
ground - based spectroscopy confirms their passive evolution and cluster membership ( dressler & gunn 1992 ) ; morphological classification comes from wide field planetary camera 2 ( wfpc2 ) imaging in the f555w and f814w bands ( smail et al .
we assume the currently popular cosmology of @xmath13 , @xmath14 , and @xmath15 km s@xmath16 mpc@xmath16 , although our results are more sensitive to @xmath5 than to the assumed cosmology .
using the stis far - uv camera on 19 aug and 21 aug 1999 , we observed a @xmath17 field centered at ra(j2000)=@xmath18 and dec(j2000)=@xmath19 in the cluster cl0016 + 16 .
the total exposure was 27892 sec ; 10 frames were taken in two visits , with dithering by 6 pixels to smooth out small - scale detector variations .
we used the crystal quartz filter ( f25qtz ) for a bandpass that spans 14502000 , thus reducing the sky background from geocoronal and lyman-@xmath20 to negligible levels at little cost in galaxy light , as the redshift ( @xmath0 ) puts the lyman limit at the short wavelength cutoff .
because the long wavelength cutoff of this bandpass is due to detector sensitivity , red leak is also negligible ; we avoid the problematic red leak and red grating scatter that hampered earlier attempts to measure the uv upturn at moderate redshift ( e.g. , windhorst et al .
the photometric calibration is reliable at the 0.15 mag level ( baum et al .
recent efforts have revised the calibration slightly ; we assume the latest revision , with an accuracy that should be better than 0.15 mag . for reference
, we assume that a flat spectrum of @xmath21 erg -0.2 in 0.1 in sec@xmath16 @xmath22 @xmath16 produces one count per sec in the stis bandpass .
a full description of the instrument can be found in woodgate et al.(1998 ) and kimble et al .
( 1998 ) .
unfortunately , the mean temperature of the far - uv camera has increased since stis commissioning , and the dark background increases as a function of temperature .
this increase appears as a `` glow '' that is strongest in the upper left - hand quadrant of the detector ( figure 1 ) , where the dark rate can be 20 times higher than the nominal 6@xmath23 cts sec@xmath16 pix@xmath16 .
the glow varies from weak to strong in our 10 frames .
the galactic foreground extinction toward our field was thought to be @xmath24 mag at the time these observations were planned , based upon maps ( burstein & heiles 1984 ) .
the subsequent release of a new extinction map ( schlegel , finkbeiner , & davis 1998 ) , based upon an iras map of the galactic dust , revises the extinction estimate to 0.057 mag . the increased dark background and higher extinction both work to decrease the sensitivity of our observations to far - uv emission , but do not preclude constraints on the uv upturn .
when accounting for this extinction , we assume the empirical model of cardelli , clayton , & mathis ( 1989 ) , which has been parameterized and tested from the uv to ir .
our stis field includes four giant e galaxies ( see figure 1 ) ; three are detected at @xmath252.5@xmath26 significance .
we detect three other objects from the smail et al .
( 1997 ) catalog at a significance of @xmath27 .
we use the bright background spiral , dg244 ( @xmath28 ; dressler & gunn 1992 ) , to confirm the registration of our frames , and to determine the positions of the fainter galaxies .
the accuracy of relative astrometry within wfpc2 or stis images is better than 0.1@xmath29 , but the accuracy of absolute astrometry can be worse than 1@xmath29 , so the positions of objects in the stis field are determined relative to this bright spiral .
most of the objects in our chosen field are too red and faint to be detected by stis ; their non - detection does not suggest an unexpected lack of sensitivity , and all detected and non - detected objects are consistent with the expected range of spectral energy distributions ( seds ) .
stis repeatedly images uv - bright stars ( e.g. , globular cluster ngc6681 ) , and no drop in sensitivity is apparent in observations before and after our observation of the cluster cl0016 + 16 .
the stis data was reduced via methods nearly identical to those used for the hubble deep field south , to which we refer the interested reader for more details ( gardner et al.1999 ) . in brief , the images were processed via the standard pipeline , excluding the dark subtraction , low - frequency flat field correction , and geometric correction .
dark frames from july through august of 1999 were processed in the same manner , and those with a strong glow ( @xmath30 cts sec@xmath16 pix@xmath16 ) were summed and fit with a cubic spline to produce a profile appropriate for our cl0016 + 16 observations ( the shape of the glow changes slowly with time , and so recent darks are required for this fit ) . a flat component and glow component to the dark background were then subtracted from each cl0016 + 16 frame , and then the flat field correction was applied .
the frames were registered by integer shifts and summed via the drizzle package .
the pixels in each frame were weighted by the ratio of the exposure time squared to the dark count variance , including a hot pixel mask .
the algorithm weights the exposures by the square of the signal - to - noise ( s / n ) ratio for sources that are fainter than the background , thus optimizing the summation to account for the temporal and spatial variations in the dark background .
the statistical errors in the final drizzled image ( cts pix@xmath16 ) , for objects below the background , are given by the square root of the final drizzled weights map scaled by the exposure time .
geometric correction , which tends to smear out the pixels through non - integer shifts , was not applied . for small dithers and large extraction apertures , it is not needed for object registration .
wfpc2 f814w data for the same field were obtained from the stsci archive and reduced via standard techniques , including cosmic - ray rejection and masking of problematic pixels .
the total exposure in the f814w image is 16800 sec .
the summed image was used to determine the restframe optical - band flux for each elliptical galaxy , and to determine the relative positions of objects in the stis field .
we performed aperture photometry on the far - uv frames using idl , taking a weighted average of the flux within a 16-pixel ( 0.4@xmath29 ) radius , and a weighted average of the flux within a sky annulus of radii 80 and 100 pixels .
the aperture includes the bright core of each galaxy , as they appear in the wfpc2 data .
the aperture is small enough to minimize the background in the far - uv measurement , but large enough to avoid significant errors in the expected encircled energy ( due to the uncertainty in the position of the galaxies ) ; it is also relevant to measurements made of local galaxies , which use a nuclear aperture ( see [ secinterp ] ) .
the weighting used the map of statistical errors ( see [ secred ] ) , in the sense that pixels were weighted less if they had less exposure ( due to masked pixels ) or higher dark count rates .
this weighting did _ not _ weigh by counts in the _ data _ frame , which would obviously bias the photometry toward pixels with more source counts .
we determined the local sky value from the mean instead of the median in the sky annulus , because most pixels are ones and zeros .
note that alternate methods for determining the sky background ( e.g. , fitting a surface to the local residual background ) yield results well within the 1@xmath26 photometric errors .
the positions of objects in the stis image were determined from the geometrically - corrected wfpc2 f814w frame , relative to the spiral galaxy dg244 ( see figure 1 ) .
calculation of position in the stis frame involves a rescaling , rotation , translation , and geometric distortion , using the geometric distortion coefficients of malumuth ( 1997 ) .
we tested our positioning algorithm using wfpc2 and stis images of the globular cluster ngc6681 , which was observed with stis numerous times before and after our cl0016 + 16 observations at various roll angles and positions .
absolute positional accuracy for objects in the stis frame is 12 stis pixels , well within the 16-pixel radius used for aperture photometry . for the wfpc2 f814w frames , we performed aperture photometry with the iraf package phot , using a 4-pixel radius ( 0.4@xmath29 ) , and a sky annulus of radii 20 and 25 pixels , thus matching the photometry done in the stis image . note that this aperture size would produce encircled energy agreement at the 5% level for point sources , and better agreement for extended sources ( robinson 1997 ; holtzman et al .
1995 ) , so the uncertainty in encircled energy contributes less than 0.1 mag to @xmath8 .
we do not perform an aperture correction because we are only interested in colors , not the absolute fluxes .
photometry was performed on all objects with @xmath31 mag in the smail et al .
( 1997 ) catalog , plus a faint object ( # 696 ) that is obvious in the stis image but faint in the wfpc2 frames .
table 1 gives the photometry for the four giant elliptical galaxies , plus three objects detected at @xmath27 in the stis image .
the uv upturn is traditionally characterized by the restframe @xmath8 color ( see burstein et al . 1988 ) .
conversion to restframe @xmath8 from our observed bandpasses depends upon the assumption of an sed .
brown et al . ( 1998 ) used the spectra of three local elliptical galaxies ( ngc1399 , m60 , and m49 ) to convert their observed foc bandpasses to restframe @xmath8 , and we used the same spectra here .
we redshifted the spectra of ngc1399 , m60 , and m49 to @xmath0 , and then applied a foreground reddening of @xmath32 mag ( schlegel et al .
we then used the iraf calcphot routine to calculate the stis and wfpc2 countrates , giving : 0.05 in @xmath33 mag , + + 2.25 in where @xmath34 is the stis countrate ( cts sec@xmath16 ) , and @xmath35 is the wfpc2 countrate ( dn sec@xmath16 ) .
this conversion gives the restframe @xmath8 values shown in table 1 .
we also calculate the @xmath8 assuming the @xmath36 upper limit to the far - uv flux .
note that the @xmath8 values in table 1 would be redder if the foreground extinction is closer to @xmath37 mag ( see [ secobs ] ) , or if the far - uv seds are dominated by post - agb stars ( see below ) .
the @xmath8 color for the sum of the giant elliptical galaxy photometry gives an average measurement of the uv upturn in our sample .
although we are looking to a significantly earlier epoch , the assumption of a model sed over an empirical sed makes little difference for the restframe @xmath38 , because the observed f814w band overlaps with restframe @xmath38 .
for example , assuming that e galaxies are @xmath25 gyr old at @xmath0 , we calculated the f814w@xmath39 color using the bruzual & charlot ( 1993 ) instantaneous burst sed of age 5 gyr , and the ngc1399 sed , assuming @xmath0 with a foreground reddening of @xmath32 .
the difference in f814w@xmath39 for the two seds was only 0.02 mag .
the assumed sed has somewhat more effect on the far - uv .
our stis bandpass spans restframe 9351290 , and the @xmath8 color is based on the average flux from 12501850 .
however , we note that no model sed has been tested in the far - uv for passively - evolving galaxies at moderate redshift , and thus we prefer empirical seds .
younger galaxies should have a larger contribution from relatively hot post - agb stars , instead of hb stars ( see brown et al .
1997 ) , so our conversion might systematically produce a bluer @xmath8 than the true restframe @xmath8 . alternatively ,
if younger galaxies are dominated by even cooler hb stars than those in local galaxies , we are calculating a redder @xmath8 than the true value .
another source of systematic error is the aperture size .
local e galaxies have been measured through 14@xmath29 diameter aperture ( 1.5 kpc at the distance of virgo ) , sampling only the nuclear flux ( burstein et al .
the @xmath4 measurements were made through a 1.82@xmath29 diameter aperture ( 9.8 kpc at the distance of abell 370 ) , including all of the galactic light detected by the foc ( brown et al .
we have measured the flux from our @xmath0 galaxies through a @xmath40 diameter aperture ( 5.4 kpc at the distance of cl0016 + 16 ) , enclosing the light from the galaxy cores .
our aperture is 3.6 times larger than that used for local galaxies , but 1.8 times smaller than that used for the abell 370 galaxies .
ohl et al .
( 1998 ) found that local e galaxies usually ( but not always ) become redder in @xmath8 at increasing radius ; the color gradient is not large ( 0.10.5 mag ) over radii 725@xmath29 , and the color of the integrated light in an increasing aperture will change even less , because most of the light comes from the central @xmath41 kpc even in our larger apertures .
thus , the aperture effects should be small , but perhaps non - negligible ( @xmath42 mag ) .
the average restframe @xmath8 color for the four giant e galaxies at @xmath0 is much redder than that observed in clusters at @xmath3 and @xmath4 . at the @xmath36 upper limit to the far - uv flux ,
none of the @xmath0 galaxies are as blue as the strong uv - upturn galaxies observed locally ( e.g. , m60 or ngc1399 ) .
ideally , one would want to compare galaxies at similar velocity dispersion ( @xmath43 ) in each epoch , because @xmath43 can be measured in a model - independent manner , and because the uv upturn strongly correlates with @xmath43 locally ( burstein et al .
measurements of @xmath43 are unavailable for some of the @xmath4 galaxies and all of the @xmath0 galaxies , but these galaxies were selected from the brightest and largest in each cluster , and would likely show strong uv upturns if observed locally .
the faint far - uv emission from the @xmath0 galaxies thus demonstrates that the uv upturn is sensitive to age .
we plot these uv upturn measurements in figure 2 , along with the models of tantalo et al.(1996 ) , which trace the chemical evolution of elliptical galaxies under the assumption of gas infall .
note that the scatter in local uv upturn measurements is affected by the wide range of sizes and luminosities in the sample ; the moderate - redshift samples are more homogeneous , but the measurements themselves are less statistically significant . in the tantalo et al .
( 1996 ) models , the uv upturn appears at an age of @xmath44 gyr , becomes quite blue by 9 gyr , and is flat thereafter . assuming the models are correct and that the oldest stars in cluster e galaxies all formed at a common redshift , our measurements imply @xmath45 .
if elliptical galaxies form through hierarchical merging , the age of the oldest stars may predate such merging ; alternatively , the age of the oldest stars may reflect the age of the galaxies themselves , if they were assembled through monolithic collapse .
however , the tantalo et al.(1996 ) models rely on several parameters ( e.g. , time of onset for galactic winds , efficiency of the star formation rate , accretion timescale , etc . ) that can be tuned to delay or accelerate the onset of the uv upturn . in this sense ,
our results say more about the evolution of hb morphology than @xmath5 ; if galaxies form their oldest stars at @xmath46 , our data imply that a blue hb population can not arise until ages greater than 7 gyr .
although the elliptical galaxies in our program are chosen through ground - based spectroscopy that confirms their passive evolution , the strong uv upturn found at @xmath4 could theoretically be due to residual star formation instead of evolved populations .
this star formation would have to cease by the present epoch , because the uv emission from local cluster elliptical galaxies is clearly due to hb stars instead of star formation .
if all clusters are alike , our new @xmath0 observations reinforce the case against star formation at @xmath4 , because such star formation would likely be even stronger at a larger lookback time .
measurements at @xmath47 0 , 0.375 , and 0.55 demonstrate that the uv upturn rises sharply over a few gyr , and then levels off .
this line of inquiry into the evolution of galaxies is still in the early stages , and further observations are needed over a range of redshifts .
such measurements should become much more feasible in hst cycle 10 , once a cooler is connected to stis ( reducing the dark background to minimal levels ) .
support for this work was provided by nasa through the stis gto team funding .
tmb acknowledges support at gsfc by nas 5 - 6499d .
we wish to thank the morphs project for making their cluster data and catalogs publicly available . | the restframe uv - to - optical flux ratio , characterizing the `` uv upturn '' phenomenon , is potentially the most sensitive tracer of age in elliptical galaxies ; models predict that it may change by orders of magnitude over the course of a few gyr . in order to trace the evolution of the uv upturn as a function of redshift
, we have used the far - uv camera on the space telescope imaging spectrograph to image the galaxy cluster cl0016 + 16 at @xmath0 .
our @xmath1 field includes four bright elliptical galaxies , spectroscopically confirmed to be passively evolving cluster members .
the weak uv emission from the galaxies in our image demonstrates that the uv upturn is weaker at a lookback time @xmath25.6 gyr earlier than our own , as compared to measurements of the uv upturn in cluster e and s0 galaxies at @xmath3 and @xmath4 .
these images are the first with sufficient depth to demonstrate the fading of the uv upturn expected at moderate redshifts .
we discuss these observations and the implications for the formation history of galaxies . |
there are quantum mechanical problems in which the hamiltonian depends explicitly on time , for example , the interaction of the system with an external time - dependent field where @xmath0 . in such cases ,
the system does not remain in any stationary state , and the behavior of it is governed by the time - dependent schrdinger equation @xmath1 the problem will now be an initial - value ( cauchy ) problem of solving this partial differential equation for a prescribed initial condition at @xmath2 . when problems of this short are discussed formally , it is common to speak of the perturbation ( @xmath3 ) as causing _ transitions _ between eigenstates of @xmath4 . if this statement is interpreted to mean that the state has changed from its inital value to a final value , than it is incorrect .
the effect of a time - dependent perturbation is to produce a nonstationary state , rather than to cause a jump from one stationary state to another which were determined by considering boundary conditions solely . experiments open the possibility to investigate dynamical properties of confined systems . in that field , of particular interest
is the response of the system to time - dependent variations of the confining field .
then , even assuming complete isolation of the many - body confined system from the environment , there is the question of how the correlation properties change
. furthermore , one can consider an atom in its ground state .
if at times @xmath5 it becomes subject to a time - varying potential energy caused by a charged heavy projectile passing by , its electron cloud will be shaken up so that its energy increases .
the energy change , related to the stopping power , is a measurable quantity .
however , when we have shake - up processes , the application of one - particle auxiliary pictures is not necessarily useful _ a priori_. motivated by such problems , in this work we consider a simplified interacting model system introduced by heisenberg @xcite in the early days of quantum mechanics @xmath6 where @xmath7}$ ] measures the strength of repulsive interparticle interaction .
we will use ideas and methods @xcite for solving the quantum motion of a particle in a harmonic oscillator with time - dependent frequency , by adding a quadrupolar ( @xmath8 ) time - dependent perturbation , of @xmath9 character , to the above ground - state hamiltonian @xmath10 in the field of heavy - particle interaction with atoms , such term could mimic the shake - up process due to close encounters . there , as was stressed @xcite by fermi , quantum mechanics
is needed since bohr s classical treatment is valid @xcite only for dipolar perturbation .
one might think that a two - particle model is a bit trivial to test time - dependent many - body methods .
but , this is not the case .
indeed , it is the few - body correlated dynamics that serve @xcite as benchmark for methods beyond , for instance , time dependent density functional theory , td - dft .
furthermore , to the theory of breathing modes of many - body systems in harmonic confinement , it was utilized @xcite to replace a given ( say , with coulombic interparticle interaction ) hamiltonian with a quadratic hamiltonian for which an analytic solution exist .
a mapping at the hamiltonian level could be a practical version of the isospectral deformation discussed earlier @xcite in the light of the , unfortunately , formal character of pure existence theorems and mapping lemmas behind td - dft .
we believe that experience about relevant details in interacting systems is best governed by studying time - dependent quantities based on exactly solvable instructive examples .
in particular , the simultaneous treatment of the notorious kinetic and interparticle energy components could be more accurate than in td - dft where both are approximated .
besides , our result allows a transparent implementation of the mapping - formalism @xcite designed to construct independent - particle potentials ( and to discuss @xcite an alarming paradox arising there ) by using exact probabilistic quantities of the correlated model as inputs . as a last motivation
, we note that there have been efforts to clarify the basic relations between interparticle correlation and information - theoretic measures for entanglement .
for instance , the stationary ground - state of the correlated model applied in the present work has already been investigated in this respect . a surprising duality between rnyi s entropies characterizing entangled systems , with attractive ( @xmath11 ) and repulsive interparticle interactions
was found @xcite and explained @xcite recently .
the question of how such a remarkable duality will change when the correlated system is perturbed by a time - dependent external field is an exciting one of broad relevance .
one of our goals here is to provide an answer .
we stress that we restrict ourselves to the linear - response limit , by taking our @xmath12 sufficiently small or sudden , since the main goal is on the time - dependent spectral aspects of correlation - dynamics and not on comparison with data .
however , we should note that similarly to the born series of stationary scattering theory , an order - by - order expansion could result in only an asymptotic series .
indeed , higher - order response to external field is an important topic in the energy change during shake - up dynamics of a nucleus @xcite or an atom @xcite . the idea that by renormalization of the kinetic energy one could @xcite extend the validity of linear - response formalism is also a challenging one .
such questions need accurate solutions in linear and nonlinear responses .
the nonlinear version of the exact determination of energy changes in our correlated model system will be published separately .
the rest of the paper is organized as follows .
section ii contains our theoretical results .
section iii is devoted to a short summary and few relevant comments .
we will use natural , rather than atomic , units in this work , except where the opposite is explicitly stated .
by introducing the normal coordinates @xmath13 and @xmath14 , one can easily rewrite @xcite the unperturbed hamiltonian in the form @xmath15 where @xmath16 and @xmath17 denote the resulting independent - mode frequencies .
it is this separated form which shows that a time - dependent perturbation of dipolar character , @xmath18\ , x_1 $ ] , will couple only to one normal mode characterized by the unperturbed angular frequency @xmath19 .
therefore , such perturbation @xcite produces an energy change independent of the correlated aspect of our model . in the case
investigated here , we have @xmath20 , i.e. , there is time - dependent quadrupolar perturbation in both independent normal modes , which will evolve in time independently .
thus , we proceed by one oscillator [ @xmath21 in a time - dependent harmonic confinement , following the established theoretical path , where one has to solve a time - dependent schrdinger equation , of the form given by eq .
( 1 ) , with @xmath22 the solution rests on making proper changes of the time and distance scales @xcite in order to consider frequency variations in @xmath23 as it changes from @xmath24 during time - evolution .
the exact mode , a nonstationary evolving state @xmath25 $ ] , contains these scales as @xmath26^{1/4}\ , \exp\left[-\frac{x^2}{2}\ , \frac{m\ , \omega_0}{\hbar \ ,
r^2(t ) } \left(1-i\ , \frac{r(t)\dot{r}(t)}{\omega_0}\right)\right]\ , e^{-i\ , \gamma(t)}.\ ] ] this exact quantum mechanical , time - dependent solution is obtained @xcite by considering a classical equation of motion @xmath27 with complex @xmath28 $ ] substitution , where the real @xmath29 gives the length scale at the instant @xmath30 .
the nonlinear differential equation , determining this scale becomes @xmath31 ^ 3},\ ] ] after taking @xmath32 .
the initial conditions are @xmath33 and @xmath34 .
since we are dealing with a weak external perturbation in this work on spectral dynamics , we solve eq .
( 7 ) via the substitutions @xmath35 and @xmath36 .
with @xmath37 to eq .
( 7 ) , we get a forced - oscillator - like differential equation @xmath38 which is treated by going to the complex notation @xmath39 $ ] in order to derive a first - order differential equation for @xmath40 .
remarkably , eq . ( 8) shows transparently that the breathing frequency @xcite is precisely @xmath41 , i.e. , the double of the confinement frequency characterizing the ground - state mode of the unperturbed @xmath42 . once an explicit form for @xmath43 is prescribed to be used in both ( @xmath44 ) differential equations for @xmath45 [ considering the two independent _ modes _ , with @xmath46 and @xmath47 and the solutions for @xmath48 are found , the time - dependent wave function becomes @xmath49 which is valid for @xmath50 , and to which the form for @xmath51 is given by eq .
( 5 ) . by using this normal - coordinate - based representation for the exact wave - function we can easily calculate with it and the _ unperturbed _ hamiltonian , the expectation values of the kinetic and potential energy components , @xmath52 and @xmath53 , respectively . we get for these quantities @xmath54\ ] ] @xmath55 in the linear - response limit , the total energy _ change _
@xmath56 becomes @xmath57}.\ ] ] to arrive at the consistent r.h.s .
above , which is valid for weak external fields , we linearized eqs .
( 10 - 11 ) . in this case
[ with eq .
( 8) ] one gets @xmath58 $ ] for the rate .
next , we derive for the overlap @xmath59 the following expression @xmath60 in terms of @xmath61 . this quantity is a measure of correlated ( when @xmath62 dynamics .
its product form reflects the independence of two normal modes during propagation .
when we are close to the stability limit at @xmath63 , even a weak external field could result in an almost perfect orthogonality at certain , @xmath43-dependent , time @xmath5 .
now , we turn to quantities which show explicitly the entangled nature of the correlated two - particle system in the time domain . by rewriting the wave function in terms of original coordinates , we determine the reduced single - particle density matrix [ @xmath64 from @xmath65 after a long , but straightforward , calculation we obtain @xmath66 where , for further clarifications below , we introduced the following abbreviations @xmath67^{1/4}\ , e^{-\frac{m}{2\hbar}\omega_s(t)\ , x^2[1-i\ , \alpha(t)/\omega_s(t)]},\ ] ] @xmath68\ ] ] @xmath69 ^ 2 + [ \dot{r}_1(\omega_1,t)/r_1(\omega_1,t ) - \dot{r}_2(\omega_2,t)/r_2(\omega_2,t)]^2 } { \omega_1(t)+\omega_2(t)}\ ] ] with @xmath70 $ ] , where @xmath71 ^ 2 $ ] .
the one - matrix is hermitian , as it must be , since @xmath72 .
later we will give , as in the stationary case @xcite earlier , a direct spectral decomposition of this important hermitian matrix in eq.(15 ) , without considering , as more usual , an eigenvalue problem .
the diagonal of the reduced single - particle density matrix gives the basic variable of td - dft , i.e. , the one - particle probability density @xmath73^{1/2}\ , e^{-\frac{m}{\hbar}\omega_s(t)\ , x^2}.\ ] ] the single - particle probability current , @xmath74 , is calculated as usual @xcite from @xmath75 which , in terms of the functions introduced above , becomes @xmath76 these probability densities satisfy the continuity equation : @xmath77 . by taking the @xmath78 substitution in eq.(15 ) ,
i.e. , neglecting the important role of the interparticle coordinate , we can define an auxiliary independent - particle model , where the particles move in a _ certain _ external field .
the wave function of this modeling becomes @xmath79 this approximate state will still result in the exact probability density and exact probability current , by construction .
so , one may consider it as an _ output _ of td - dft . indeed , using eqs .
( 6.50 - 6.52 ) of @xcite or eqs .
( 21 - 22 ) of @xcite , an effective potential @xmath80 is given by @xmath81\ , = \
, \frac{1}{2}\omega_{s}^2(t)\ , x^2\ , -\ , \frac{1}{2}\left[\sqrt{\omega_{s}(t)}\ , \frac{d^2}{d t^2}\left(\frac{1 } { \sqrt{\omega_{s}(t)}}\right)\right ] x^2,\ ] ]
at least upto a time - dependent constant @xcite . in the so - called adiabatic treatment in td - dft ,
one uses @xcite only the first term as a time - dependent external field .
but , with such a single - particle potential , which produces from a time - dependent wave equation the @xmath82 state , we also arrive at a dilemma .
we know the exact schrdinger hamiltonian .
especially , we know how the unperturbed hamiltonian is modified by adding a time - dependent quadrupolar perturbation to it . furthermore , since the perturbation acts only over a limited time in our problem , in the schrdinger picture we can follow the standard quantum mechanical prescription to determine the energy change by calculating expectation values with the exact wave function @xmath83 and unperturbed hamiltonian @xmath84 . in the light of @xmath85 $ ] , the situation in the orbital version of td - dft is not so clear .
for instance , it will contain a @xmath86-proportional higher - order derivative @xcite . quite unfortunately , in td - dft , which is based @xcite on a density - potential mapping lemma , we do not @xcite know universal functionals to determine an energy change via them .
after the above remarks on open questions in applied dt - dft , and motivated partly by dreizler s early suggestion @xcite on using an isospectral deformation ( @xmath87 ) and a constrained search , we turn to the decomposition of our reduced one - particle density matrix , and , in addition , put forward an idea on the possibility of using an _ approximate _
nonidempotent one - matrix , @xmath88 , designed below .
following our experience @xcite on the decomposition of the stationary @xmath89 , we will use mehler s formula @xcite now to @xmath90 .
thus , via the @xmath91=\bar{\omega}(t)[1+z^2(t)]/[1-z^2(t)]$ ] and @xmath92 $ ] correspondences , we introduce two new variable @xmath93 and @xmath94 .
the solutions are @xmath95 @xmath96 - \sqrt{\left[\frac{\omega_s(t)}{a(t)}+1\right]^2 - 1}.\ ] ] now , in term of these we get our point - wise decomposition directly @xmath97[z(t)]^l\ , \phi_l(x_1,t)\ , \phi_l^{*}(x_2,t),\ ] ] where the @xmath98 elements of the orthonormal basis set ( natural orbitals ) are @xmath99^{1/4}\ , e^{i\ , \frac{m}{2\hbar}\alpha(t)x^2}\ , \frac{1}{\sqrt{2^l\ , l!}}\ , e^{-\frac{m}{2\hbar}\bar{\omega}(t)x^2}\ , h_m(\sqrt{m\bar{\omega}(t)/\hbar}x).\ ] ] of course , extracted quantities , like the exact one - matrix @xmath90 and the associated exact probability density @xmath100 , contain less and less information then the wave function @xmath101 and the pair - function @xmath102 .
however , even by these extracted quantities one can calculate the exact kinetic , and the external - potential energies , respectively , while in td - dft only the last energy is , in principle , exact .
thus , to formulate an idea , which is based on inversion from _ known _ probability density @xmath100 and probability current @xmath74 ( i.e. , from the crucial @xcite @xmath103 phase ) , we make a rewriting @xmath104[z(t)]^l [ 1-z_d(t)][z_d(t)]^l|\phi_l[x , t,\omega_d(t)]|^{2},\ ] ] which is valid if and only if @xmath105/[1-z_d(t)]$ ] .
so , with one - particle probabilistic functions as input , we can prescribe , from the r.h.s .
of the above equation , a trace - conserving @xmath88 which has the form of @xmath90 .
such nonidempotent form contains two ( interconnected ) parameters @xmath106 and @xmath107 , instead of the uniquely derived @xmath93 and @xmath94 .
whether this deformed one - matrix could result in , because of its _ tunable _ flexibility , an essentially better approximation for the kinetic energy than the td - dft [ @xmath108 where the one - matrix is idempotent , requires future investigations within the framework of a constrained search @xcite .
the information - theoretic aspects of iterparticle interaction , i.e. , of a crucial term in the hamiltonian , may be considered via entropic measures which do not use expectation values with physical dimensions but use pure probabilities encoded in the occupation numbers @xmath109\ , [ z(t)]^l},\ ] ] which are , in the present case , time - dependent quantities as well .
first , we calculate rnyi s entropy @xcite , with @xmath110 and @xmath111 , from @xmath112^q}{1-[z(t)]^q}}.\ ] ] the von neumann entropy is obtained as a limiting case when @xmath113 .
it is given by @xmath114\ , -\frac{z(t)}{1-z(t)}\ , \ln[z(t)]}.\ ] ] in the knowledge of @xmath94 , these entropies characterize the deviation from the stationary case described by @xmath115 , i.e. , without external perturbation on the correlated two - body system . by considering repulsive [ @xmath116 and attractive [ @xmath11 ] versions , we can investigate the duality @xcite problem with respect to entropies , now in the time domain .
notice , that the knowledge of rnyi entropies for arbitrary values of @xmath118 provides , mainly from information - theoretic point of view , more information than the von neumann entropy alone , since one could extract @xcite from them the full spectrum of the one - matrix .
but , as eq .
( 26 ) shows transparently , to physical expectation values one also needs the proper basis set .
armed the above , strongly interrelated , theoretical details on the dynamics of interparticle correlation we come to their implementation .
we use a simple model for illustration , with @xmath119 . by tuning the value of the effective rate of change ( denoted by @xmath120 ) we can easily investigate the sudden and adiabatic limits .
we get @xmath121\ ] ] @xmath122\ ] ] @xmath123\ ] ] the sum ( @xmath44 ) of @xmath124 measures the total deviation from the unperturbed ground - state energy @xmath125 . in the kick - limit , where @xmath126 , we have @xmath127}$ ] .
as expected on physical grounds , the change at long time , i.e. , taking @xmath128 first , becomes a constant which tends to zero in the @xmath129 sudden limit . in the long - time limit the overlap parameter of eq .
( 13 ) , which measures the time - evolution of orthogonality between exact states , also tends to a constant value .
simple independent - particle modelings of the correlated model system could be based on effective ( @xmath130 ) , prefixed external fields @xmath131 before time - dependent perturbation .
in such cases one gets @xmath132}}$ ] for the total energy change , following the standard quantum mechanical averaging procedure discussed at eq .
two options are analyzed here , and a remarkable result is found , as figure 1 signals .
namely , with @xmath133 , i.e. , with the density - optimal modeling , the corresponding energy _ change _ tends to the exact one at high enough values of @xmath120 , since @xmath134 .
somewhat surprisingly , by using the so - called energy - optimal hartree - fock @xcite modeling , where @xmath135 , one can not get agreement with the exact result at any value of @xmath120 . for illustration , a convenient ratio - function is defined as @xmath136 and plotted in figure 1 as a function of the @xmath120 with fixed , density - optimal , @xmath137 .
we have checked , using both sides of eq .
( 12 ) , that with a @xmath138 value we are in the linear - response limit , where the energy changes are proportional to @xmath139 .
notice , that we use standard hartree atomic units , where @xmath140 , to both figures of this work .
[ 0.3 ] , as a function of @xmath30 which is measured in atomic units .
the solid and dashed curves refer to interparticle repulsion ( @xmath141 ) and attraction ( @xmath142 ) , respectively .
we used @xmath138 and @xmath143 .
the top , middle , and bottom panels refer , respectively , to the @xmath144 , @xmath145 , and @xmath146 values .
[ figure1],title="fig : " ] as we can see from the figure , far from the sudden limit , i.e. , at a small @xmath120 value , the density - optimal modeling underestimates the true energy change in the correlated case .
we speculate that such sensitivity close to the adiabatic limit could be , at least partially , behind yet unclarified controversies in the applied td - dft @xcite to energy transfer by slow ions in wide - gap insulators , where @xmath147 is positive of course . in section iii
, we will return to the informations given in figure 1 , considering there the energy changes for negative @xmath147 . at the end of this section
, we turn to illustrations of the spectral aspect of dynamic interparticle correlation .
we apply von neumann entropy to get a precise insight into the time - dependence of an information - theoretic measure . in the light of facts on hamiltonian - based measures ,
the energy - change and the wave - function overlap at @xmath128 , such analysis is desirable .
based on it , one can discuss , and maybe understood , challenging interrelations between measures which are based on different emphasizes within quantum mechanics .
we stress here that the exact energy - changes depend , as expected , on the sign of @xmath147 .
[ 0.3 ] , as a function of the time @xmath30 , which is measured in atomic units .
the solid and dashed curves refer to a moderate interparticle repulsion @xmath148 , and the corresponding ( see the text ) interparticle attraction @xmath149 , respectively .
three values for the rate parameter @xmath120 , measured in inverse time - units , are employed to illustration .
top panel : @xmath150 .
middle panel : @xmath151 .
bottom panel : @xmath152 .
[ figure2],title="fig : " ] in figure 2 we exhibit @xmath153 , by taking @xmath143 , @xmath138 , @xmath154 and @xmath142 .
these couplings for repulsion and attraction , respectively , are fixed to yield equal entropies at @xmath155 , in order to shed further light on duality @xcite , now in the time - domain as well .
notice , that in the unperturbed case to _ any _ allowed [ @xmath156 $ ] repulsive coupling there exists a corresponding attractive ( @xmath11 ) one for which the calculated entropies are equal . for instance , when we are close to the stability limit in the repulsive ( @xmath157 ) case , i.e. , when @xmath158 , the corresponding attractive ( @xmath159 ) coupling becomes @xmath160 .
the solid and dashed curves refer to repulsion and attraction , as we mentioned .
they are plotted as a function of the time measured in atomic units .
the rate - parameter of the time - dependent external perturbation is described by @xmath120 and measured in inverse units of time . in order to discuss tendencies
, we used three values for this parameter . as expected , at very short times the entropies , similarly to the energy changes ,
grow in time .
after about 1 - 2 atomic units ( i.e. , about 25 - 50 attoseconds ) they start to reflect the oscillating character builded into the time - dependent occupation numbers @xmath161 , the primary root of which is in the time - evolving correlated wave function .
the entropies will keep , in absence of an environment - made dissipative @xcite coupling , such oscillating behavior without limitation . from this point of view , they signal the time - evolvement of the closed interacting system .
more importantly , one can see from the illustration that the initial dual character of entropies , fixed at @xmath155 , will _ disappear _ due to a new physical variable , i.e. , the time .
finally , there are reductions in the oscillation - amplitudes by increasing @xmath120 . clearly , at the so - called sudden limit ( @xmath129 ) for a time - dependent perturbation
, they will diminish as expected , since the system s reaction - rapidity is limited by its normal mode frequencies . in order to justify the above argument more directly , we take now a somewhat less realistic modeling by using an abrupt quench of _ all _ interactions at @xmath162 in the hamiltonian . in that case , instead of eqs .
( 33 - 34 ) , one gets @xcite the following exact expressions @xmath163 @xmath164 there is no change in the system s kinetic energy since @xmath165 ^ 2+[\dot{r}_i(\omega_i , t)/\omega_i]^2=1 $ ] . by using time - dependent @xmath166 and @xmath167 , which are needed to eq .
( 25 ) , we get @xmath168 for all @xmath169 since @xmath170\equiv{2[2\sqrt{\omega_1\omega_2}/(\omega_1-\omega_2)]^2}$ ] .
thus , there is no entropy-_change _ in the total - quench case .
this @xmath171 character could be a useful constraint in practical attempts , with a good starting @xmath100 , using the interconnected @xmath107 and @xmath106 variables of eq .
one can take @xmath172 , where @xmath173 .
based on the time - dependent schrdinger equation , an exact calculation is performed for the time - evolving energy change , and orthogonality of initial and final states , in a correlated two - particle model system driven by a weak external field of quadrupolar character . besides , considering the structure of the exact time - dependent one - matrix , an independent - particle model is defined from it which contains exact information on the single - particle probability density and probability current .
the resulting noninteracting auxiliary state is used to construct an effective potential and discuss its applicability .
we analyzed the energy changes in a comparative manner , and pointed out ( see , below as well ) the limited capability of an optimized independent - particle modeling for the underlying time - dependent process .
a point - wise decomposition of the reduced one - particle density matrix is given in terms of time - dependent occupation numbers and time - dependent orthonormal orbitals .
based on such exact spectral decomposition on the time domain , an entropic measure of inseparable correlation is also investigated .
it is found that the duality behavior characterizing the stationary ground - state may disappear depending on the rapidity of the external time - dependent perturbation . at realistic rapidity , the undamped time - dependent oscillations , found in an entropic measure , reflect the fact that the correlated system evolves in time . in this respect
, we have a signal on the excited state , similarly as one can see in the time - evolving probability density and probability current .
our first comment is based on results obtained for the energy changes , and plotted in figure 1 , by considering the sign of @xmath147 . in the case of an attractive interparticle interaction ,
the case of a nucleus @xcite , the exact result for @xmath174 becomes bigger at moderate @xmath120 values than the result obtained within the density - optimal framework @xmath175 .
so , we speculate that a recent result @xcite on the excitation of a deformed nucleus within td - dft could have the same , underestimating character in its quadrupole channel , as the atomistic case has .
thus , the true enhancement in excitations over the result based on the modeling of teller @xcite , might also need further theoretical refinement for that channel . in the second comment , we focus on the role of the switching rate @xmath120 in the case of energy changes .
as a preliminary step to future realistic calculations , here we would like to estimate a characteristic interaction - time ( @xmath176 ) in classical atom - atom collision . to do that , we model the three - dimensional screened interaction via a finite - range ( @xmath177 ) potential @xmath178 solving the classical problem @xcite for the collision time ( @xmath179 ) we get @xmath180\ ] ] where @xmath181 is the energy of a heavy charged ( @xmath182 ) projectile moving with velocity @xmath183 and colliding at impact parameter @xmath184 with a fixed center of _ effective _ charge @xmath185 .
the dimensionless parameter introduced is @xmath186 $ ] , i.e. , it is related to the ratio of potential and kinetic energies .
a closer inspection of this compact ( and reasonable at short - range solid - state conditions ) form shows that practically one may use @xmath187 to get a very acceptable estimation as a function of the heavy projectile velocity @xmath183 , with which @xmath188 is _ decreasing _ from 4 ( at @xmath189 ) to 2 ( at @xmath190 ) . if we take @xmath191 and consider the energy change determined above , we get a @xmath192^{-2}}\sim{[\alpha(v)v]^2}$ ] character at very low velocities . for the opposite
, high - velocity , limit @xmath192^{-2}}\sim{[\alpha(v)v]^{-2}}$ ] is the scaling , which also looks quite reasonable physically .
however , the detailed work on the exactly determined non - linear energy changes is left for a dedicated publication . there , and
also in bohr s pioneering modeling @xcite with time - dependent dipole fields of a passing charge , the time - scale derived above using collision theory , could play a more quantitative role to understand fine details behind experimental predictions @xcite on energy losses of slow ions in insulators .
furthermore , in the renewed field of time - dependent energy losses , a proper time - scale also could help to analyze further a remarkable theoretical prediction @xcite on the strong interplay of nuclear and electronic stopping components of the observable total energy transfer .
indeed , few - electron shake - up processes can be dominant at close encounters in condensed matter . in such dynamical cases , one - electron approximations , with double - occupancy for an auxiliary spatial orbital , could result in inaccuracies .
we thank professor p.
m. echenique for the very warm hospitality at the dipc .
one of us ( in ) is grateful to professor p. bauer and professor d. snchez - portal for useful discussions on energy transfer processes in insulators .
this work was supported in part by the spanish ministry of economy and competitiveness mineco ( project no . fis2013 - 48286-c2 - 1-p ) .
00 w. heisenberg , z. phys . *
38 * , 411 ( 1926 ) .
m. moshinsky , am . j. phys .
* 36 * , 52 ( 1968 ) .
l. e. ballentine , _ quantum mechanics _ ( world scientific , singapore , 1998 ) . v. s. popov and a. m. perelomov , sov .
jetp * 30 * , 910 ( 1970 ) . yu .
kagan , e. l. surkov , and g. v. shlyapnikov , phys .
a * 54 * , r1753 ( 1996 ) .
a. del campo , phys .
* 111 * , 100502 ( 2013 ) .
e. fermi , phys . rev .
* 57 * , 485 ( 1940 ) .
n. bohr , phil . mag . * 25 * , 10 ( 1913 ) .
j. f. dobson , phys .
rev . lett . * 73 * , 2244 ( 1994 ) .
m. brics and d. bauer , phys . rev .
a * 88 * , 052514 ( 2013 ) . c. r. mcdonald , g. orlando , j. w. abraham , d. hochstuhl , m. bonitz , and t. brabec , phys .
lett . * 111 * , 256801 ( 2013 ) .
h. kohl and r. m. dreizler , phys .
. lett . * 56 * , 1993 ( 1986 ) . c. a. ullrich , _ time - dependent density - functional theory _ ( oxford university press , oxford , 2012 ) , and references therein .
m. ruggenthaler , m. penz , and r. van leeuwen , arxiv : 1412.7052v1 [ cond-mat.other ] .
j. schirmer and a. dreuw , phys . rev .
a * 75 * , 022513 ( 2007 ) .
j. pipek and i. nagy , phys .
a * 79 * , 052501 ( 2009 ) .
schilling , d. gross , and m. christandl , phys .
lett . * 110 * , 040404 ( 2013 ) . m. l. glasser and i. nagy , phys .
a * 377 * , 2317 ( 2013 ) .
schilling and r. schilling , j. phys .
a : math . theor .
a * 47 * , 415305 ( 2014 ) .
d. w. robinson , nucl .
phys . * 25 * , 459 ( 1961 ) .
k. w. hill and e. merzbacher , phys .
rev . a * 9 * , 156 ( 1974 ) .
m. mierzejewski and p. prelovsek , phys .
. lett . * 105 * , 186405 ( 2010 ) .
m. lein and s. kmmel , phys .
lett . * 94 * , 143003 ( 2005 ) .
i. nagy , phys .
a * 87 * , 052512 ( 2013 ) .
a. erdlyi , _ higher transcendental functions _ ( mcgraw - hill , new york , 1953 ) , p. 194 .
p. koscik , phys .
a * 379 * , 293 ( 2015 ) .
j. rapp , m. brics , and d. bauer , phys .
a * 90 * , 012518 ( 2014 ) . c. l. benavides - riveros and i. nagy , arxiv : 1406.2809v1 [ quant - ph ] .
a. rnyi , _ probability theory _
( north - holland , amsterdam , 1970 ) .
p. calabrese , p. le doussal , and s. n. majumdar , phys .
rev . a * 91 * , 012303 ( 2015 ) . m. ashan zeb , j. kohanoff , d. snchez - portal , and e. artacho , nucl .
methods phys .
b * 303 * , 59 ( 2013 ) .
f. mao , ch .
zhang , j. dai , and r .- s .
zhang , phys .
rev . a * 89 * , 022707 ( 2014 ) .
k. albrecht , phys .
56b * , 127 ( 1975 ) , and references therein .
i. v. tokatly , phys .
110 * , 233001 ( 2013 ) .
i. stetcu , c. a. bertulani , a. bulgac , p. magierski , and k. j. roche , phys .
lett . * 114 * , 012701 ( 2015 ) .
m. goldhaber and e. teller , phys .
rev . * 74 * , 1046 ( 1948 ) .
i. nagy , nucl .
methods phys .
b * 94 * , 377 ( 1994 ) .
s. p. moller , a. csete , t. ichioka , h. knudsen , u. i. uggerhoj , and h. h. anderson , phys .
lett . * 93 * , 042502 ( 2004 ) , and references therein .
s. n. markin , d. primetzhofer , and p. bauer , phys .
. lett . * 103 * , 113201 ( 2009 ) .
a. a. correa , j. kohanoff , e. artacho , d. snchez - portal , and a. caro , phys .
* 108 * , 213201 ( 2012 ) , and references therein . | time - dependent quantities are calculated in the linear response limit for a correlated one dimensional model atom driven by an external quadrupolar time - dependent field . besides the analysis of the time - evolving energy change in the correlated two - particle system , and orthogonality of initial and final states , mehler s formula is applied in order to derive a point - wise decomposition of the time - dependent one - matrix in terms of time - dependent occupation numbers and time - dependent orthonormal , natural orbitals .
based on such exact spectral decomposition on the time domain , rnyi s entropy is also investigated .
considering the structure of the exact time - dependent one - matrix , an independent - particle model is defined from it which contains exact information on the single - particle probability density and probability current .
the resulting noninteracting auxiliary state is used to construct an effective potential and discuss its applicability . |
experimental results from flavor physics and electroweak precision measurements have long established the standard model pattern of flavor and cp violation .
the only sources of flavor and cp violation are the yukawa couplings , and the one essentially unknown parameter is the relative coupling strength of the heavy third - generation quark to the @xmath0 boson , the ckm mixing angle @xmath1 @xcite .
its knowledge is crucial to establish the unitarity of the ckm mixing matrix in the standard model .
because the su(3 ) symmetry of qcd does not see electroweak charges , this coupling can not be determined in qcd - mediated top pair production .
instead , we rely on the electroweak production process for a single top quark in association with a quark jet to measure this parameter of the standard model .
the problem of modern particle physics is that while on the one hand we have good reasons to expect that we will see a non - trivial ultraviolet completion of the standard model at the tev scale , we do not know where in such an extended model this particular flavor structure originates from .
one of the standard candidates for such new physics at the tev scale with all its benefits from a dark matter candidate and stabilized higgs mass to a valid grand unified theory is the minimal supersymmetric standard model mssm @xcite .
mainly in the soft - breaking terms in the mssm lagrangian there are multiple sources of flavor and cp violation which naturally predict observable effects , including flavor changing neutral currents . assuming only one set of supersymmetric partner states we can implement the experimental constraints by postulating a symmetry dubbed minimal flavor violation .
this implies that there still be no sources of flavor violation other than the yukawa couplings , the spurions of flavor symmetry breaking @xcite
. an interesting alternative might be the promotion of the gauge sector to @xmath2 supersymmetry , leading to dirac fermion partners of the standard model gauge bosons and additional scalar particles , all clearly visible at the lhc @xcite . in this paper ,
we instead focus on the more subtle effects of an approximate minimal flavor violation symmetry .
focussing on the quark / squark sector , minimal flavor violation forces the soft squark masses to be almost diagonal in flavor space and the scalar trilinear @xmath3 terms for example describing squark - squark - higgs interactions to be proportional to the yukawa couplings .
corrections consistent with the standard model flavor symmetry are induced by higher powers in the yukawa couplings .
since we can not derive minimal flavor violation from first principles , we need to measure if and by how much it is broken . in the down - quark sector
, squarks contribute to @xmath4-and @xmath5-physics observables via squark - gluino loops mediated by the strong coupling constant @xmath6 , which gives us powerful tests of minimal flavor violation .
in contrast , in the the up - quark sector such one - loop effects are proportional to the weak coupling @xmath7 or the yukawa couplings and much harder to measure . the first - third and second third generation mixing between the @xmath8 and @xmath9 with @xmath10 squarks is essentially invisible for kaon , charm and b experiments @xcite .
integrating out the heavy supersymmetric particles , such loop contributions lead to flavor - violating quark couplings to standard model gauge bosons , for example a @xmath11-@xmath12-@xmath13 or @xmath14-@xmath12-@xmath13 vertex @xcite . at the lhc processes involving valence
quarks are generally more interesting , so we will focus on the @xmath11-@xmath12-@xmath13 vertex , but its second generation counter part can of course be treated the same way
. the search for this largely unconstrained effective coupling is linked with the measurement of @xmath1 through the relevant lhc processes . while the effective gluon vertex leads to the direct production of an isolated top quark ,
single top production is accompanied by a quark jet .
however , at the lhc we know that the radiation of additional quark and gluon jets from the incoming quarks is ubiquitous . therefore , the question becomes : how can we tell apart electroweak ckm effects in single top production ( and its two production mechanisms ) and strong effects from non - minimal new physics in direct top production , all including top quark decays as well as realistic qcd effects and backgrounds .
in spite of the fact that there are many flavor observables constraining squark mixing beyond minimal flavor violation , there are two entries in the squark mass matrices which are still largely unconstrained . even though we actually use the diagonalized squark mass matrix , it is instructive to discuss the results in terms of dimensionless mass insertions @xmath15 ( for the squark handedness @xmath16 , the generation indices @xmath17 , and the weak isospin @xmath18 ) .
the @xmath19 are the relevant off - diagonal entries and @xmath20 the corresponding mean diagonal entry in the squark mass matrix . because we are not interested in the strongly constrained down - type mixings @xmath21 and we focus on left - right mixing only
, we denote @xmath22 the unconstrained left - right mixing terms are the off - diagonal @xmath23 between @xmath8 and @xmath10 and the diagonal @xmath24 @xcite .
the left - right swapped @xmath25 instead mixes @xmath26 and @xmath27 .
it is constrained by @xmath28 transitions @xcite and @xmath29 in @xmath30 mixing @xcite .
the reason why the bounds on @xmath25 are strong and those on @xmath23 hardly exist is the chargino - top loop : if the chargino is a mix of the wino and the higgsino , the latter will have a large yukawa coupling to the external bottom .
the @xmath31 instead couples to the wino content of the chargino , which forces is to be left handed @xmath26 .
this @xmath25 constraint can only be relaxed by heavy squarks .
assuming minimal flavor violation the generation - diagonal entry @xmath32 is of the general form @xmath33 and does not need to be small . as a matter of fact
, it can lead to a large splitting of the two stop masses and ameliorates the little hierarchy problem , so we do not expect it to be small either . it is currently only constrained via the lower limit on the light higgs mass and can be measured either in stop mixing or in the minimal supersymmetric higgs sector @xcite .
the generation mixing entry @xmath23 mixes @xmath8 and @xmath10 and at one - loop induces the flavor - changing chromo - magnetic operator @xcite @xmath34 via a squark - gluino loop . as shown on the left fig .
[ fig : feyn_dtop ] this operator implies direct top production at hadron colliders @xmath35 , with a leptonic @xmath0 decay to avoid an undetectable purely hadronic final state .
the corresponding partonic cross section is typically suppressed by the heavy gluino mass in the loop and therefore proportional to @xmath36 the second diagram in fig.[fig : feyn_dtop ] contributes proportional to @xmath37 , which is negligible after fulfilling all flavor constraints . due to the absolute values
squared they are invariant under complex phases of the squark mixing parameters .
this process to our knowledge is the only way to measure the otherwise unaccessible @xmath23 in the era of lhc and super-@xmath5 factories .
usually , the mssm parameter space beyond minimal flavor violation is huge , and physical processes typically involve many possible contributions which are free to cancel each other .
in contrast , direct top production is a strongly interacting process and depends only on the masses @xmath38 and @xmath39 and the squark mixing parameters @xmath40,@xmath23 and @xmath25 . for small values of @xmath41
we have to include @xmath42 in the expression for @xmath24 . for our numerical study
we use the sps2 @xcite inspired reference point with the gut - scale masses @xmath43 gev , @xmath44 gev and @xmath45 .
the higgs sector is characterized by @xmath46 and @xmath47 .
the relevant weak - scale masses for our process are @xmath48 this gives us a light ( mostly ) stop mass of 955 gev and a light higgs mass of 117 gev .
the lightest neutralino is a viable dark matter candidate . for the calculation of the direct top production rate at this point we use feynarts , formcalc @xcite , looptools @xcite , and hadcalc @xcite .
we include all supersymmetric qcd and electroweak contributions .
there are four major backgrounds for direct top production @xmath49 with a charge identified lepton from the @xmath0 decay and a tagged bottom jet with a tagging probability of 50% and a mis - tagging probability of 1% @xmath50 of the three @xmath0 plus one jet backgrounds the irreducible combination is suppressed with respect to the mis - tagging background by roughly two orders of magnitude . similarly , the more likely mis - tag of a charm requires a @xmath51 initial state and should be at maximum of the same order as the valence - quark induced @xmath52 process .
most importantly , the kinematic distributions of the three are very similar , so we expect all three to vanish as a roughly constant fraction of the leading @xmath52 background .
the two - bottoms background process can become dangerous when the two bottom jet are close enough to look like one bottom jet from gluon splitting . if we require the two bottom jets to be close ( @xmath53 ) the resulting @xmath54 rate is of the same order as the subleading @xmath55 production , which again means that in the following discussion we will focus on the mis - tagged @xmath56 background .
all of the background and the signal pass the acceptance ( and triggering ) cuts @xmath57 without a major reduction .
note that we apply bottom acceptance cuts to the mis - tagged light jet in the background , so at this level there will not be any light - flavor jets in the analysis .
+ jets , after acceptance cuts .
these distributions ( and only these ) are generated without qcd jet radiation but including momentum smearing to account for detector effects.,title="fig:",scaledwidth=35.0% ] + jets , after acceptance cuts .
these distributions ( and only these ) are generated without qcd jet radiation but including momentum smearing to account for detector effects.,title="fig:",scaledwidth=35.0% ] there are two key distributions to separate direct top production from the continuum backgrounds , shown in fig .
[ fig : dtop_rec ] : due to the signal kinematics , the transverse momentum of the bottom jet is strongly peaked around @xmath58 , even including detector effects . for the signal
, the partonic center of mass energy @xmath59 should also peak around the top mass .
the unmeasured longitudinal neutrino momentum we compute from the mass shell condition @xmath60 .
the remaining two - fold ambiguity we resolve by choosing the better top mass reconstruction @xcite .
these kinematic features we exploit by requiring @xmath61 as shown in table [ tab : dtop ] the statistical significance after cutting on the kinematic features of the signal is not sufficient to extract single top production at the lhc , even though the gaussian significance for @xmath62 of data looks promising . theory and other systematic errors require a reasonable value of at least @xmath63 for such a counting experiment , which means we have to search for additional ways to extract direct top production out of the qcd background .
.signal and background rates for direct top production after acceptance cuts eq.([eq : acc_cuts ] ) , resonance cuts eq.([eq : rec_cuts ] ) , existing additional qcd jets eq.([eq : jets_cuts ] ) , and finally correlation cuts on this qcd jet activity eq.([eq : qcd_cuts ] ) .
the statistical significances assume @xmath64 of luminosity at 14 tev .
for the merged sample we only consider the leading background . [
cols="<,>,>,>,>,>",options="header " , ] + a distinguishing feature of direct top production which has nothing to do with the decay kinematics is that coming from a @xmath65 initial state the produced top quark will be boosted longitudinally following the direction of the valence quark .
we can follow this behavior by noticing that for the signal the lepton and in particular the bottom quark are moving into the forward direction , peaked around pseudorapidities of two .
in contrast , the qcd background behaves like drell - yan @xmath0 production with one parton splitting in the initial state , where the @xmath0 boson as well as the lepton are central .
interestingly enough , the radiated mis - tagged jet above our @xmath66 threshold is also fairly central .
the problem with this general kinematic feature is that it is not strong enough to allow for an efficient cut in our analysis .
this changes once we include further qcd jet radiation . to simulate the radiation of qcd jets beyond the leading jet we employ the mlm matching scheme as implemented in madevent @xcite , which allows us to consistently add @xmath67 samples with an arbitrary number of additional jets @xcite . for large transverse momenta
these jets will be correctly described by the hard matrix element and for small transverse momenta also correctly by the parton shower .
while the main motivation for such jet merging simulations are for example @xmath0+jets samples as backgrounds to higgs and new physics searches , the same method allows us to simulate jet radiation accompanying heavy states at the lhc and exploit these patterns to improve the signal extraction @xcite .
first , we ask for additional jet from qcd radiation which have to pass the staggered jet acceptance cuts @xmath68 one of these three jets should be the bottom jet from the top decay .
usually , the top decay jet will be the hardest , maybe the second hardest jet .
this means that one of the two 30 gev jets will have to in addition pass the top resonance cut of @xmath69 as listed in eq.([eq : rec_cuts ] ) .
this feature as well as our aim to be not too dependent on the details of the jet merging simulation motivates us to even for the light - flavor jets only consider maximum pseudorapidities of 2.5 . in this aspect , a more dedicated analysis of the bottom tag
could significantly improve the signal efficiency and hence the gaussian significance .
while 20 gev for the third jet sounds very small , this analysis is also meant to show the impact the analysis of the qcd jet activity can have on new - physics searches , so we decide to keep it as low as possible . depending on the structure of the measured underlying event this threshold might have to be increased eventually .
the improvement of the direct top analysis just through requiring the existence of two jets we show in table .
[ tab : dtop ] . the higher apparent probability to radiate one additional jet from the signal process
is related to a larger number of soft jets for the continuum background , as we will see in fig .
[ fig : dtop_j2 ] .
this is due to the continuum structure of the background diagrams without a hardly radiating resonant top quark and the fact that the mis - tagged bottom is actually a massless jet and more likely to split collinearly .
.,title="fig:",scaledwidth=30.0% ] .,title="fig:",scaledwidth=30.0% ] .,title="fig:",scaledwidth=30.0% ] just based on our argument above and on color factors , in the signal case the radiation off the incoming gluon will dominate the radiation pattern .
this means that the top quark and the leading qcd jet fly into opposite directions .
we can see this behavior in the pseudorapidity distributions as well as in the opening angle distributions in fig .
[ fig : dtop_j1 ] : the first radiated jet for the signal is indeed widely separated from the lepton as well as the bottom jet , from the top quark .
in contrast , the continuum qcd background radiates jets over a wide pseudorapidity range , bounded only by the maximum pseudorapidity of 2.5 .
because the lepton is central , this means that the pseudorapidity difference between the lepton and the first jet has to drop off fast once we go to pseudorapidities of order two .
the same is true for the pseudorapidity difference of the first two radiated jets , once of which is falsely tagged as a bottom jet : if both of them are reasonably central their pseudorapidity difference is rarely going to exceed values of 2.5 .
these well distinguishable jet distributions we can now cut on , to separate signal and background .
symmetrically , we constrain the two pseudorapidity differences to be @xmath70 and show the results in table [ tab : dtop ] . while these cuts do not improve the statistical significance much beyond the top reconstruction cuts they bring down @xmath71 to a manageable level .
refining these jet cuts we can further improve @xmath71 , but an the expense of the number of signal events left in the analysis .
cl signal , assuming gaussian statistics with @xmath72 .
the white area in the upper right corner is already excluded by experimental squark searches .
we adopt a squark mass bound of @xmath73.,scaledwidth=40.0% ] just out of curiosity we can check what happens once we include a second qcd jet ( altogether three jets ) in our analysis . obviously , this is not suitable for the full analysis , but it could give interesting information for those signal nd background events in which we see such an additional jet .
the distribution of the number of jets we show in the first panel of fig .
[ fig : dtop_j2 ] . in the jet distributions
the general pattern of the first qcd jet is still present largely because it is based on the boosted nature of the top quark in the hard signal process . in fig . [
fig : dtop_j2 ] we see a similar correlation between the second qcd jet and the lepton as we see for the first jet .
the pseudorapidity difference between the two qcd jets is also more strongly peaked in the generally central continuum background .
based on table [ tab : dtop ] we can compute the @xmath74 confidence level coverage of direct top production in the @xmath75 plane . in fig .
[ fig : dtop_coverage ] we see that with a mild dependence on @xmath76 the lhc will be able to rule out non - minimal flavor violation through the mass matrix entry above @xmath77 , dependent on the acquired luminosity .
vertex with a solid circle.,scaledwidth=90.0% ] at the lhc , more signal processes contribute to the one - top - plus - jets final state @xcite .
the two irreducible cousins of direct top production with jet radiation are the single top production channels shown in fig .
[ fig : feyn_stop ] : one of them proceeds via a time - like @xmath12-channel @xmath0 boson @xcite and the other through a space - like @xmath78-channel @xmath0 boson @xcite .
the associated @xmath79 production @xcite we do not consider in this analysis because its final state is significantly different and neither related to qcd not irreducible from direct top production . encouraged by the signal vs background direct top analysis in the last section we can ask if more generally the structure of qcd jet radiation @xcite will allow us to distinguish these three single top and direct top production mechanisms and tell apart the responsible coupling in or beyond the standard model @xcite . usually , the three direct / single top production channels are distinguished using bottom tags , where aside from the top decay products the @xmath12-channel process involves a forward bottom jet while the @xmath78-channel process is accompanied by a central bottom jet and direct to production does not involve additional bottom jets at all @xcite . already for
the single top channels alone , identifying the different qcd features first and cross checking for bottom flavor later might allow us to improve the @xmath1 measurement in and beyond the standard model . studying the forward @xmath80 jet in the @xmath12-channel process without tagging it from the beginning also improves our sensitivity to flavor changing @xmath81-@xmath12-@xmath0 couplings enhanced by the valence quark parton densities @xcite .
we simulate direct top production as well as the two single top production processes including two additional hard qcd jets .
additional qcd jets can arise from the parton shower and are described over their entire phase space by mlm jet merging implemented in madevent @xcite .
all @xmath82 events appearing as part of the merged sample we subtract . because at this stage we are not suggesting a dedicated analysis we first assume we know which of the jets are bottom jets and which are light - flavor jets . for all of them we require
the staggered acceptance cuts @xmath83 the pseudorapidity range is the same for bottom and light - flavor jets and avoids distinguishing bottoms and light - flavor jets .
this way we can later apply our analysis to all - flavor jets . to ensure that these signal events pass the triggers we request one lepton from the top decay with @xmath84 without showing all the distributions we know what to expect for the @xmath66 spectrum of the accompanying jets in the three processes : in the @xmath12-channel process there will be one hard central jet balancing the top quark and a second forward @xmath80 jet from the gluon splitting . tagging this forward bottom jet might or might not be useful , dependent for example on the optimization of the signal vs background analysis with respect to @xmath71 or @xmath72 @xcite .
this forward bottom jet will be collinearly divergent , regularized by the bottom mass , down to transverse momenta of @xmath85 its @xmath66 spectrum will diverge . the qcd jet balancing the top peaks at transverse momenta around @xmath86 , almost as high as the bottom jet from the top decay ( with its jacobian peak around 65 gev ) .
this we can understand when we consider this process as one - sided weak boson fusion .
dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] + dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] + dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] + dependence as discussed in the text .
the label @xmath87 represents all - flavor jets which might be bottom jets ( @xmath80 ) or light - flavor jets ( @xmath88).,title="fig:",scaledwidth=30.0% ] + in the @xmath78 channel the flavor structure of the standard model enforces a second bottom jet from the off - shell @xmath0 decay .
this jet plays a similar balancing role as the light - flavor jet in the @xmath12-channel process .
we now we see two competing hard bottom jets of comparable @xmath66 , both peaking around @xmath89 .
direct top production in contrast does not predict any additional hard jets in the detector . because it involves a flavor changing production vertex
there should be no additional bottom jets , and the light - flavor jets will follow the typical initial state radiation pattern . to describe the angular correlations of the final state @xcite including qcd jets in more detail we consider the observable @xmath90 .
it parameterizes the angle between @xmath91 in the rest frame of @xmath92 and @xmath93 in the lab frame .
it is not symmetric in its arguments ; if the two particles are back to back and @xmath94 it becomes @xmath95 . in between
it vanishes in the case where @xmath91 in the center of mass frame is orthogonal to the lab frame movement of this center of mass .
this behavior we confirm in the first row of fig .
[ fig : stop_j1 ] . for @xmath12-channel single top production the hardest light - flavor jet balances the top ,
so the lepton is most likely back to back with the hardest jet .
because it balances the heavy top , the leading jet is harder than the lepton and the @xmath96 distribution peaks at @xmath95 . for the @xmath78 channel
the decay lepton will be central , as will be the first qcd jet .
except for the azimuthal angle we expect no back - to - back configuration , which gives us @xmath97 values flat around zero . in direct top production
the first jet is radiated at high rapidity . because of the color factor it most likely comes from the gluon , which means it will be collinear as well as soft .
the top is boosted against the incoming gluon and with it the decay lepton .
this means that the lepton and the first qcd jet are back to back , and the hard decay lepton will be more energetic than the soft collinear jet .
the @xmath97 distribution will then peak around @xmath98 .
considering further jets in @xmath12-channel single top production the lower energy of subsequent jet radiation washes out the @xmath99 behavior .
this is due to the role of the energy hierarchy with respect to the lepton in this observable .
once we arrive at jet number three all we see is a general parton shower jet radiation , slightly forward but uncorrelated with the top decay products . for the direct top channel
the pattern of the second jet resembles the first , because the radiation off the incoming gluon will still dominate .
similarly , additional jets accompanying @xmath78-channel top production will become slightly more forward and hence more likely to move towards larger @xmath100 values , but without an exploitable structure . .
we also show the correlation of the leading bottom jet with the first two light - flavor jets.,title="fig:",scaledwidth=30.0% ] .
we also show the correlation of the leading bottom jet with the first two light - flavor jets.,title="fig:",scaledwidth=30.0% ] .
we also show the correlation of the leading bottom jet with the first two light - flavor jets.,title="fig:",scaledwidth=30.0% ] + .
we also show the correlation of the leading bottom jet with the first two light - flavor jets.,title="fig:",scaledwidth=30.0% ] .
we also show the correlation of the leading bottom jet with the first two light - flavor jets.,title="fig:",scaledwidth=30.0% ] + following our original motivation , what is most interesting are angular correlation of light - flavor and bottom jets without assuming a @xmath80 tag . in the right column of fig [ fig : stop_j1 ]
we see that for @xmath12-channel single top production the hardest all - flavor jet indeed balances the top quark , it is not bottom flavored .
the second hardest jet then comes from the top decay , which we can check by comparing it features with the bottom correlations shown in fig .
[ fig : stop_b1 ] .
the third and forth jets both arise from parton splitting and have not distinct correlations with the top decay products being bottom flavored or not . checking their @xmath66 distribution
we can confirm that they simultaneously drop off fast at values above 40 gev .
for @xmath78-channel single top production both leading all - flavor jets have little to do with the @xmath96 distribution , which means they are the two bottom jets from the hard process .
again , we confirm this behavior in fig .
[ fig : stop_b1 ] . in this figure
we also see that essentially all bottom jets prefer values @xmath101 , which means the bottom jets balance the lepton direction and the bottom energy lies above the lepton energies .
this is simply an effect of the intermediate @xmath0 decay step which softens the @xmath0 decay product as compared to the bottom jet . for direct top production
the hardest jet is usually the bottom decay jet .
in addition we expect no more bottom jets , so the distributions of @xmath102 match those of @xmath103 .
because the argument of the color factor preferring radiation off the incoming gluon combined with the boosted center of mass frame holds in general , subsequent jets show a similar preference for @xmath104 . beyond the individual jet - lepton correlations we can also study the correlations between the different all - flavor jets . in fig .
[ fig : stop_2d ] we should sample distributions of this kind for the three production processes under consideration . for the @xmath12 channel process we already know that the two leading jets , one light - flavor and one the bottom decay jet , both reside at small values of @xmath105 .
the third and fourth jet , again one bottom and one light - flavor , do not have a strong preference for large values of @xmath106 , but in contrast to the leading two jets they show a correlation with each other , based on the general radiation pattern . in the @xmath78-channel process
the two leading jets are the two bottom jets .
they are correlated and prefer small values of @xmath107 , dependent mostly on the lepton energy . for the next two jets the correlations is considerably weaker , and from fig .
[ fig : stop_j1 ] we already know not to expect much in formation in their @xmath105 distributions .
direct top production produces a hard bottom jet with a slight preference towards @xmath108 . aside from that
, the universal jet radiation structure shows a diagonal correlation with the slightly favored region @xmath104 , becoming much more prominent for the two subleading jets . for all - flavor jets . from left to
right we show @xmath12-channel single top production , @xmath78-channel single top production , and direct top production.,title="fig:",scaledwidth=30.0% ] for all - flavor jets . from left to right we show @xmath12-channel single top production , @xmath78-channel single top production , and direct top production.,title="fig:",scaledwidth=30.0% ] for all - flavor jets . from left to right we show @xmath12-channel single top production , @xmath78-channel single top production , and direct top production.,title="fig:",scaledwidth=30.0% ] + for all - flavor jets . from left to right we show @xmath12-channel single top production , @xmath78-channel single top production , and direct top production.,title="fig:",scaledwidth=30.0% ] for all - flavor jets . from left to right we show @xmath12-channel single top production , @xmath78-channel single top production , and direct top production.,title="fig:",scaledwidth=30.0% ] for all - flavor jets . from left
to right we show @xmath12-channel single top production , @xmath78-channel single top production , and direct top production.,title="fig:",scaledwidth=30.0% ]
in this paper we have shown how we can exploit the qcd jet activity to extract direct top production from the qcd background and to tell apart @xmath12-channel single top production , @xmath78-channel single top production , and direct top production .
direct top production is the only way to observe or constrain the supersymmetric flavor - violating squark mass entry @xmath109 in the near future .
it is mediated by the flavor changing neutral current @xmath11-@xmath12-@xmath13 interaction and has to be extracted from @xmath0+jets backgrounds with a light - flavor jet mis - tagged as a bottom .
while kinematic cuts based on the top resonance structure of the matrix element are not sufficient to produce a promising signal - to - background ratio @xmath71 , we find that additional cuts on the jet correlation from qcd radiation improve this ratio enough to allow for a meaningful lhc analysis .
the same kind of qcd jet correlations with the top decay lepton allow us to distinguish ( normalized ) samples of @xmath12-channel single top production , @xmath78-channel single top production , and direct top production .
all three channels are irreducible if we consider searches for one top quark plus jets .
our distinction is purely based on angular correlations and serves as an orthogonal test to possible bottom tags . in single
top searches applying explicit bottom tags only later in the analysis allows us to make use of the parton density enhancement when looking for flavor - changing top couplings .
both methods combined should be able to unambiguously determine the third generation flavor structure of the standard model , including the ckm mixing element @xmath1 as well as new - physics effects .
we thank christoph englert , stefan gieseke , christoph hackstein for many helpful discussions and tim tait for teaching us single tops and carefully reading the manuscript .
johan alwall we thank for help with madevent .
our work was supported by the deutsche forschungsgemeinschaft via the sonderforschungsbereich / transregio sfb / tr9 `` computational particle physics '' . c. amsler _ et al . _
[ particle data group ] , phys .
b * 667 * , 1 ( 2008 ) ; f. abe _ et al .
_ [ cdf collaboration ] , phys . rev .
lett . * 74 * , 2626 ( 1995 ) .
s. abachi _ et al . _
[ d0 collaboration ] , phys .
lett . * 74 * , 2632 ( 1995 ) . v.
m. abazov _ et al . _
[ d0 collaboration ] , phys .
d * 78 * , 012005 ( 2008 ) .
h. e. haber and g. l. kane , phys .
rept . * 117 * , 75 ( 1985 ) .
g. dambrosio , g. f. giudice , g. isidori and a. strumia , nucl .
b * 645 * , 155 ( 2002 ) ; b. c. allanach , g. hiller , d. r. t. jones and p. slavich , jhep * 0904 * , 088 ( 2009 ) .
g. d. kribs , e. poppitz and n. weiner , phys .
d * 78 * , 055010 ( 2008 ) ; g. d. kribs , a. martin and t. s. roy , arxiv:0901.4105 [ hep - ph ] . t. plehn and t. m. p. tait , j. phys .
g * 36 * , 075001 ( 2009 ) .
l. j. hall , v. a. kostelecky and s. raby , nucl .
b * 267 * , 415 ( 1986 ) ; j. s. hagelin , s. kelley and t. tanaka , nucl .
b * 415 * , 293 ( 1994 ) ; f. gabbiani , e. gabrielli , a. masiero and l. silvestrini , nucl . phys .
b * 477 * , 321 ( 1996 ) ; m. misiak , s. pokorski and j. rosiek , adv .
high energy phys . * 15 * , 795 ( 1998 ) .
m. ciuchini , e. franco , a. masiero and l. silvestrini , phys .
d * 67 * , 075016 ( 2003 ) [ erratum - ibid .
d * 68 * , 079901 ( 2003 ) ] ; a. bartl , k. hidaka , k. hohenwarter - sodek , t. kernreiter , w. majerotto and w. porod , arxiv:0905.0132 [ hep - ph ] .
s. dittmaier , g. hiller , t. plehn and m. spannowsky , phys .
d * 77 * , 115001 ( 2008 ) .
t. m. p. tait and c. p. yuan , phys .
d * 55 * , 7300 ( 1997 ) ; j. cao , g. eilam , k. i. hikasa and j. m. yang , phys . rev .
d * 74 * , 031701 ( 2006 ) ; j. guasch , w. hollik , s. penaranda and j. sola , nucl .
suppl . * 157 * , 152 ( 2006 ) ; j. j. cao , g. eilam , m. frank , k. hikasa , g. l. liu , i. turan and j. m. yang , phys .
d * 75 * , 075021 ( 2007 ) .
j. a. casas and s. dimopoulos , phys
. lett .
b * 387 * , 107 ( 1996 ) .
_ et al . _ ,
lett . * 96 * , 221601 ( 2006 ) ; b. aubert _ et al . _ [ babar collaboration ] , arxiv : hep - ex/0612017 .
e. barberio
_ et al . _
[ heavy flavor averaging group ( hfag ) ] , arxiv : hep - ex/0603003 . http://www.slac.stanford.edu/xorg/hfag r. lafaye , t. plehn , m. rauch and d. zerwas , eur .
j. c * 54 * , 617 ( 2008 ) ; r. lafaye , t. plehn , m. rauch , d. zerwas and m. duhrssen , arxiv:0904.3866 [ hep - ph ] .
e. malkawi and t. m. p. tait , phys .
d * 54 * , 5758 ( 1996 ) ; m. hosch , k. whisnant and b. l. young , phys .
d * 56 * , 5725 ( 1997 ) ; t. han , m. hosch , k. whisnant , b. l. young and x. zhang , phys . rev .
d * 58 * , 073008 ( 1998 ) .
a. belyaev , arxiv : hep - ph/0007058 .
b. c. allanach _ et al .
_ , eur .
j. c * 25 * , 113 ( 2002 ) .
t. hahn , comput .
commun . * 140 * , 418 ( 2001 ) ; t. hahn and c. schappacher , comput .
commun . * 143 * , 54 ( 2002 ) .
t. hahn and m. perez - victoria , comput .
. commun . * 118 * , 153 ( 1999 ) .
m. rauch , arxiv:0804.2428 [ hep - ph ] .
m. l. mangano , m. moretti and r. pittau , nucl .
b * 632 * , 343 ( 2002 ) .
j. alwall _ et al .
_ , jhep * 0709 * , 028 ( 2007 ) ; j. alwall , p. artoisenet , s. de visscher , c. duhr , r. frederix , m. herquet and o. mattelaer , aip conf . proc . * 1078 * , 84 ( 2009 ) .
s. catani , f. krauss , r. kuhn and b. r. webber , jhep * 0111 * , 063 ( 2001 ) .
j. alwall , s. de visscher and f. maltoni , jhep * 0902 * , 017 ( 2009 ) ; j. alwall , k. hiramatsu , m. m. nojiri and y. shimizu , arxiv:0905.1201 [ hep - ph ] .
d. o. carlson , arxiv : hep - ph/9508278 ; z. sullivan , phys .
d * 70 * , 114012 ( 2004 ) . for a new review see w. bernreuther , j. phys .
g * 35 * , 083001 ( 2008 ) .
s. s. d. willenbrock and d. a. dicus , phys . rev .
d * 34 * , 155 ( 1986 ) ; s. dawson and s. s. d. willenbrock , nucl .
b * 284 * , 449 ( 1987 ) ; d. o. carlson and c. p. yuan , phys .
b * 306 * , 386 ( 1993 ) ; t. stelzer , z. sullivan and s. willenbrock , phys .
d * 56 * , 5919 ( 1997 ) ; j. m. campbell , r. k. ellis and f. tramontano , phys . rev .
d * 70 * , 094012 ( 2004 ) ; q. h. cao , r. schwienhorst , j. a. benitez , r. brock and c. p. yuan , phys .
d * 72 * , 094027 ( 2005 ) .
s. frixione , e. laenen , p. motylinski and b. r. webber , jhep * 0603 * , 092 ( 2006 ) .
b. w. harris , e. laenen , l. phaf , z. sullivan and s. weinzierl , phys .
d * 66 * , 054024 ( 2002 ) ; q. h. cao , r. schwienhorst and c. p. yuan , phys .
d * 71 * , 054023 ( 2005 ) .
t. m. p. tait , phys .
d * 61 * , 034001 ( 2000 ) ; s. zhu , arxiv : hep - ph/0109269 ; j. m. campbell and f. tramontano , nucl .
b * 726 * , 109 ( 2005 ) ; s. frixione , e. laenen , p. motylinski , b. r. webber and c. d. white , jhep * 0807 * , 029 ( 2008 ) .
r. d. peccei , s. peris and x. zhang , nucl .
b * 349 * , 305 ( 1991 ) ; t. m. p. tait and c. p. p. yuan , phys .
d * 63 * , 014018 ( 2001) q. h. cao , j. wudka and c. p. yuan , phys .
b * 658 * , 50 ( 2007 ) ; j. a. aguilar - saavedra , nucl .
b * 804 * , 160 ( 2008 ) .
a. heinson , a. s. belyaev and e. e. boos , phys .
d * 56 * , 3114 ( 1997 ) ; t. stelzer , z. sullivan and s. willenbrock , phys .
d * 58 * , 094021 ( 1998 ) ; a. s. belyaev , e. e. boos and l. v. dudko , phys .
d * 59 * , 075001 ( 1999 ) .
f. maltoni , z. sullivan and s. willenbrock , phys .
d * 67 * , 093005 ( 2003 ) ; e. boos and t. plehn , phys . rev .
d * 69 * , 094005 ( 2004 ) .
j. m. campbell , r. frederix , f. maltoni and f. tramontano , arxiv:0903.0005 [ hep - ph ] .
z. sullivan , phys .
d * 72 * , 094034 ( 2005 ) ; g. mahlon , arxiv : hep - ph/0011349 ; p. motylinski , arxiv:0905.4754 [ hep - ph ] . | top plus jets production at hadron collider allows us to study the couplings of the top quark . in the standard model ,
two single top processes contribute to the top - jets final state . beyond the standard model ,
additional direct top production can occur .
all three processes probe top gauge couplings including flavor mixing .
the structure of accompanying qcd jets allows us to separate the direct top signal from the qcd backgrounds as well as to disentangle the three top plus jets production mechanisms orthogonally to the usual bottom tags . |
investigating deuterium chemistry is useful to put constraints on the ionization fraction , temperature , density and thermal history of dense molecular clouds @xcite .
the observations of multiply deuterated molecules in space , e.g. @xcite and references therein , have shown the necessity to reexamine some reaction rates in chemical networks @xcite , elemental d / h ratio in cold dense gas @xcite and the density structure in sources such as l1544 and @xmath11 oph d @xcite , as well as the effects of accretion on grains @xcite , possible effects of internal dynamical motion @xcite , and the evolution of ice mantles in dense clouds and cores @xcite .
the first multiply deuterated interstellar molecule detected has been d@xmath1co almost twenty years ago @xcite . since then
, the study of deuterated molecules in the ism has rapidly increased as they have been proven to be a unique observational probe of the early stages in low - mass star formation .
multiply deuterated species such as triply deuterated ammonia have been detected with a surprisingly high abundance ratio of 10@xmath12 with respect to their fully hydrogenated forms @xcite . by comparing this ratio with the elemental d / h ratio ( 1.65@xmath13 , linsky et al .
1993 ) it is easily seen that there is a remarkable enrichment in deuterium in this and other molecules . in iras
16293 - 2422 ceccarelli et al .
( 1998 ) measured d@xmath1co / h@xmath1co = 5% . in the same source parise
( 2004 ) measured cd@xmath0oh / ch@xmath0oh = 1.4% .
this enrichment in deuterium has mainly been explained by the exothermicity of the h - d exchange reactions , and by the depletion of co and o onto dust grains @xcite , which enhances the abundance of the multiply deuterated forms of h@xmath14 and , upon dissociative recombination , the d / h ratio .
roberts et al .
( 2003 ) calculated d / h abundance ratios close to 0.3 ( 20,000 times the cosmic value ) , in regions with large amount of co freeze - out . despite extensive studies of nearly 30 deuterated molecules in the interstellar gas , the processes regulating their formation
are not completely understood , in particular the relative importance of gas - phase versus grain - surface chemistry .
for example , simple gas phase chemical models do not reproduce the deuterium enrichment observed in some of these species .
in particular , the formation of methanol on grain mantles is presented as an explanation for its enhanced deuteration @xcite ; for h@xmath1co instead , the gas - phase reaction path can not be excluded .
in fact , in the orion bar photodissociation region hdco seems to be formed solely in the gas phase @xcite .
deuteration of ammonia can be reproduced with gas - phase models ( e.g. roueff et al .
2005 ) , although nh@xmath0 and its deuterated forms are also expected to form on the surface @xcite .
the first detection of c - c@xmath0h@xmath1 in the laboratory @xcite allowed the identification of several u - lines previously detected in space @xcite , namely the strong lines at 85338 and 18343 mhz .
since then , cyclopropenylidene has been proven to be one of the most abundant and widespread molecules in our galaxy .
it has been observed in the diffuse gas , cold dark clouds , giant molecular clouds , photodissociation regions , circumstellar envelopes , and planetary nebulae @xcite .
given the high abundance of the normal species , both c - c@xmath0hd and the singly substituted @xmath9c species ( off axis ) have been observed with a good signal to noise in cold dark clouds .
furthermore cyclopropenylidene shows an enhancement in deuterium fractionation in cold dark clouds , for example gerin et al .
( 1987 ) measured a 1:5 ratio of the 2@xmath6 - 1@xmath15 lines of c - c@xmath0hd and c - c@xmath0h@xmath1 in tmc1 .
the reactions which lead to such a high deuteration are still poorly understood .
some rates of reactions which may be involved in the formation of deuterated c - c@xmath0h@xmath1 have been measured by savi et al .
( 2005 ) but to our knowledge they have not been included in models so far .
last year the centimeter and millimeter wavelength spectra of doubly deuterated c - c@xmath0h@xmath1 have been measured in the laboratory @xcite , allowing for the first time a search for c - c@xmath0d@xmath1 in space .
like its fully hydrogenated counterpart , c - c@xmath0d@xmath1 presents the spectrum of an oblate asymmetric top with b - type transitions .
furthermore c - c@xmath0d@xmath1 shows a deuterium quadrupole splitting resolvable at very low @xmath16 .
given the presence of two equivalent off axis bosons , it has _
ortho _ and _ para _ symmetry species with relative statistical weight of 2:1 . unlike several of the six known multiply deuterated species observed in the radio band ( d@xmath1h@xmath17 , chd@xmath1oh , nhd@xmath1 , d@xmath1co , d@xmath1s , and d@xmath1cs )
, c - c@xmath0h@xmath1 is believed to form solely by gas phase reactions @xcite .
the interplay between the gas phase and grain surface reactions in the deuteration of interstellar molecules is not clear so far , partially because there are not many probes available for testing the models : c - c@xmath0h@xmath1 is an ideal molecule for this purpose because of its easily observable transitions and because it has the possibility of double deuteration . assuming c - c@xmath0d@xmath1 is formed in the gas - phase like its fully hydrogenated counterpart
, cyclopropenylidene will be a unique probe for the deuteration processes happening in the gas - phase .
furthermore , since c - c@xmath0h@xmath1 is , in terms of cloud evolution , an early - type
molecule @xcite , it is a particular useful tool to investigate early stages of a molecular cloud .
this makes observations of its deuterated forms particularly important to test time - dependent chemical codes which include deuteration processes .
here we report on the positive detection of three emission lines of c - c@xmath0d@xmath1 in the 3 mm band , namely the @xmath2 , @xmath18 , and @xmath4 transitions , towards tmc-1c and l1544 : to our knowledge this is the first search for doubly deuterated cyclopropenylidene undertaken.d@xmath1 towards l1527 in the framework of the nobeyama 45 m telescope survey at 3 mm . ]
the observations have been carried out from 2012 september 28 until october 2 at the iram 30 m telescope , located in pico veleta ( spain ) , towards the starless cores tmc-1c and l1544 .
the choice of the sources has been made on the basis of two simple criteria : the abundance of the normal species ( c - c@xmath0h@xmath1 ) and a high deuterium fractionation .
both sources are in the taurus molecular cloud , one of the closest dark cloud systems and low - mass star forming regions in our galaxy .
l1544 is a perfect test bed to investigate the initial conditions of protostellar collapse : its structure is consistent with a contracting bonnor - ebert sphere , with central densities of about 10@xmath19 @xmath20 and a peak infall velocity of @xmath210.1 km s@xmath22 at about 1000 au from the center ( e.g. keto & caselli 2010 ) .
its centrally concentrated structure and measured kinematics suggests that this is a pre - stellar core at a late stage of evolution , toward star formation .
tmc-1c is a relatively young core , with evidence of accreting material towards a core and immersed in a cloud with densities higher than those surrounding the l1544 core @xcite .
the coordinates that were used are @xmath23 = 04@xmath2441@[email protected] @xmath27 = + 25@xmath2849@[email protected] for tmc-1c , and @xmath23 = 05@xmath2404@[email protected] @xmath27 = 25@xmath2810@[email protected] for l1544 . in the case of tmc-1c
these are the same coordinates as reported by bell et al .
( 1988 ) and gerin et al .
( 1987 ) , while in the case of l1544 they correspond to the coordinates of the peak of the 1.3 mm continuum dust emission from ward - thompson et al .
( 1999 ) . in both cores we observed two lines of c - c@xmath0h@xmath1 , one line of c - h@xmath9cc@xmath1h ( off axis ) , two lines of c - c@xmath0hd and three lines of c - c@xmath0d@xmath1 , using three different tuning settings .
a summary of the observed lines is reported in table 1 .
the emir receivers in the e090 configuration were employed , and observations were performed in a frequency switching mode with a throw of @xmath31 mhz .
all four emir sub - bands were connected to the fts spectrometer set to high resolution mode ; this delivered a final spectrum with 50 khz channel spacing ( corresponding to 0.15 km s@xmath22 at 3 mm ) and a total of 7.2 ghz of spectral coverage ( nominal bandpass of 1.8 ghz per sub - band ) .
telescope pointing was checked every two hours on jupiter and was found accurate to 3 - 4 arcsec .
lines of the isotopologues of c - c@xmath0h@xmath1 listed in table 1 have been detected in both sources with very high signal to noise ratio .
a selection of spectra of c - c@xmath0h@xmath1 and isotopologues in tmc-1c and l1544 is shown in figure 1 .
table 1 lists the observed line parameters .
even the weakest line of the doubly deuterated species , 2@xmath32 - 1@xmath33 at 108 ghz , is detected at a 7.5@xmath10 level in tmc-1c ( t@xmath34 = 2.5 mk ) , and at a 9@xmath10 level in l1544 ( t@xmath34 = 4.6 mk ) .
the gildas software @xcite was employed for the data processing : high order polynomials had to be used for baseline subtraction given the strong baseline produced by the frequency switching observing mode .
the column densities and optical depths given in table 1 were calculated using the expressions given in the appendix . as was already pointed out by bell et al .
( 1988 ) , c - c@xmath0h@xmath1 shows two velocity components towards tmc-1c , one more intense at 6 km s@xmath22 and one less intense at 5.4 km s@xmath22 , see figure 1 .
there is a hint of detection of the component at 5.4 km s@xmath22 also in c - h@xmath9cc@xmath1h , but no clear presence in the deuterated species .
assuming that all lines have the same excitation temperature in both components , we expect for the component at 5.4 km s@xmath22 a line intensity of 0.06 k for c - c@xmath0hd ( 3@xmath35 - 2@xmath36 ) and 0.013 k for c - c@xmath0d@xmath1 ( 3@xmath37 - 2@xmath38 ) . comparing these estimates with the noise level in our spectra ( 0.007 k for c - c@xmath0hd and 0.002 k for c - c@xmath0d@xmath1 )
we can say that the lower velocity component is absent in the deuterated species of cyclopropenylidene : this behavior may suggest that the lower velocity component traces a hotter region , where the deuterated molecules are not present in detectable amounts . the observed line intensity ratios of c - c@xmath0d@xmath1 pose the question whether local thermal equilibrium is a valid assumption for its excitation . only for _
para_-c@xmath0d@xmath1 more than one optically thin line was detected .
the ratios of the integrated 3@xmath5 - 2@xmath6/2@xmath39 - 1@xmath40 lines should be 4.3 and 3.3 for tmc-1c ( t@xmath41 = 7 k ) and l1544 ( t@xmath41 = 5 k ) , respectively , assuming thermalization of both lines to their assumed excitation temperature , while they are 1.5 and 1.35 . to gain insight into the excitation of the lines we performed radiative transfer calculations with radex @xcite .
collision rates of c@xmath0h@xmath1 with h@xmath1 calculated by chandra et al .
( 2000 ) and supplied by the lamda database @xcite were used together with molecular constants of _ para_-c@xmath0d@xmath1 .
calculations were done on a grid in density of molecular hydrogen , n(h@xmath1 ) , and column density of _
para_-c@xmath0d@xmath1 , n(_para_-c@xmath0d@xmath1 ) , and the results are shown in figure 2 .
it is in principle possible to read the column density of c - c@xmath0d@xmath1 from figure 2 , by knowing the density of molecular hydrogen .
the nominal observed line ratios would , particulary for l1544 , result in very high column densities . the lines would be very subthermally excited and very optically thick .
however , given the fact that the observed line strengths run parallel to each other , a small change in the observed value would result in a substantial change in the ratio .
the line intensities and @xmath42 contours of both lines show that the line strengths are very close to each other along a diagonal line spanning a large range of densities and column densities ( filling factors smaller than unity would move this line to the right in the plot ) , so that much smaller values of column densities and opacities are also in agreement with the data .
the predicted excitation temperatures tend to be rather low , in the range 33.5 k , which introduces a considerable uncertainty in the column density determinations .
since the transitions used for the excitation calculation for c - c@xmath0hd and c - c@xmath0d@xmath1 have similar upper state energy , the excitation behavior will naturally be degenerated .
therefore , in the absence of data for more transitions for c - c@xmath0d@xmath1 , c - c@xmath0hd or c - h@xmath9cc@xmath1h , it is difficult to derive conclusive results . despite these uncertainties ,
we have assumed local thermodynamic equilibrium with @xmath43 values as described in the appendix .
in addition to the observation of two lines of c - c@xmath0h@xmath1 , one line of c - h@xmath9cc@xmath1h ( with the @xmath9c off of the molecular axis ) , and two lines of c - c@xmath0hd , we also claim the detection of three lines of c - c@xmath0d@xmath1 in both tmc-1c and l1544 .
this first interstellar detection of c - c@xmath0d@xmath1 is validated by the following reasons : * we detected all favorable transitions of c - c@xmath0d@xmath1 available in the covered frequency range .
* the rest frequencies employed have laboratory accuracy @xcite , and in both sources the line shapes and velocities are in agreement with each other and with those observed for more abundant isotopologues ( see figure 1 ) . * the intensities of the c - c@xmath0d@xmath1 lines are consistent with what is expected from the deuteration of the ring in these sources , i.e. c - c@xmath0h@xmath1/c - c@xmath0hd is consistent with c - c@xmath0hd / c - c@xmath0d@xmath1 , as will be discussed below . in table 2 we present the relative abundances of the deuterated species with respect to the hydrogenated ones for cyclopropenylidene over the 27@xmath30 beam for tmc-1c and l1544 obtained from this work , and also for h@xmath1co , hco@xmath17 , n@xmath1h@xmath17 and nh@xmath0 obtained from previous work . the abundance of doubly deuterated cyclopropenylidene with respect to the normal species is ( 0.4 - 0.8)% in tmc-1c and ( 1.2 - 2.1)% in l1544 .
this interval has been determined considering the differences in @xmath44 obtained from different lines .
the deuteration of c - c@xmath0h@xmath1 follows the same trend observed for other molecules in both sources .
it is interesting to note that the ratios [ c - c@xmath0d@xmath1]/[c - c@xmath0hd ] and [ c - c@xmath0hd]/[c - c@xmath0h@xmath1 ] are quite similar in both sources .
we calculated the d / h ratio of c - c@xmath0h@xmath1 in prestellar cores using the network model of aikawa et al .
( 2012 ) . for the physical structure of the core , we adopt the collapsing core model of aikawa et al .
( 2005 ; the @xmath46=1.1 model ) and also a static model of l1544 from keto & caselli ( 2010 ) . for co depletion factors consistent with those of the two objects , i.e. @xmath47 = 3.8 for tmc-1c @xcite and @xmath47 = 14 for l1544 @xcite ,
the calculated column density ratio of c - c@xmath0d@xmath1/c - c@xmath0h@xmath1 is @xmath48 , consistent with the observed value of 0.6% for tmc-1c and 1.5% for l1544 .
there is no need for any deuterium fractionation reactions of c - c@xmath0h@xmath1 on grain surfaces to account for the observed d / h ratio : the deuteration of cyclopropenylidene can be explained solely by gas - phase reactions .
the main route of formation of deuterated cyclopropenylidene is the successive deuteration of the main species via reaction with h@xmath1d@xmath17 , d@xmath1h@xmath17 , and d@xmath14 .
an example of the reaction scheme is sketched in figure 3 , considering only h@xmath1d@xmath17 as reaction partner .
the depicted cycle of reactions starts with c - c@xmath0h@xmath1 and h@xmath1d@xmath17 , producing in the first step c - c@xmath0hd and subsequently c - c@xmath0d@xmath1 .
the same reactions happen with d@xmath1h@xmath17 and d@xmath14 .
the overall process is a series of two reactions : the proton - deuteron transfer ( slow step , red arrows ) , and the subsequent dissociative recombination with electrons ( fast step , blue arrows ) .
the presence of this deuteration cycle results in a time dependent deuterium fractionation . assuming low levels of deuteration at the start
, it is expected that this level increases as a function of time , reaching a stationary level after some time .
other deuteration processes , e.g. the formation of c - c@xmath0hd from the reaction of c@xmath0h@xmath17 with hd , were found to be negligible .
the d / h ratio of cyclopropenylidene is , therefore , directly related to that of h@xmath14 , the main deuterium donor in dark interstellar clouds .
recently huang & lee ( 2011 ) have calculated highly accurate spectroscopic constants for @xmath9c and d isotopologues of c - c@xmath0h@xmath14 in order to guide the laboratory and astronomical search .
since these species are intermediates in the formation of isotopic species of c - c@xmath0h@xmath1 , their detection would be useful to put more constraints on the models .
doubly deuterated cyclopropenylidene appears to be a very interesting probe for the earliest stages of star formation .
its formation mechanism puts important constraints on gas - phase deuteration models , and suggests the possibility of using c - c@xmath0d@xmath1 as a chemical clock .
furthermore the brightness of the @xmath49 line of c - c@xmath0d@xmath1 at 97 ghz in l1544 would allow to make an on the fly map ( otf ) of the core in a reasonable amount of time , for the first time for a doubly deuterated molecule . during our observations , the line was observed with a s / n of more than 6 after just 30 minutes .
emission from c@xmath50s is spread over 10 arcmin@xmath51 in l1544 @xcite .
we estimated that an otf map of 6 arcmin@xmath51 with a noise level of 20 mk at the iram 30 m telescope with the emir receivers would be possible in 14 hours , allowing to map at the same time c - c@xmath0h@xmath1 and c - c@xmath0d@xmath1 . by mapping the core
it will be possible to locate the deuteration peak , and put more constraints on current gas - grains models .
_ acknowledgement _ + the authors thank the referee for the useful comments .
this work has been supported by sfb956 .
s. spezzano has been supported in her research with a stipend from the international max - planck research school ( imprs ) for astronomy and astrophysics at the universities of bonn and cologne .
s. schlemmer and s.b .
acknowledge support by the deutsche forschungsgemeinschaft ( dfg ) through project br 4287/1 - 1 .
l.b . acknowledges support from the science and technology foundation ( fct , portugal ) through the fellowship sfrh / bpd/62966/2009 , and he is grateful to the sfb956 for the travel allowance .
the column densities and optical depths given in table 1 were calculated using the following expressions .
the line center opacity @xmath52 is where k is the boltzmann constant , @xmath56 is the frequency of the line , @xmath57 is the planck constant , @xmath58 is the speed of light , @xmath59 is the einstein coefficient of the transition , @xmath60 is the full width at half maximum , @xmath61 is the degeneracy of the upper state , @xmath62 is the energy of the upper state , @xmath63 is the partition function of the molecule at the given temperature @xmath43 .
@xmath43 , @xmath64 , @xmath65 are the excitation , the background ( 2.7 k ) and the main beam temperatures respectively , in k. to calculate @xmath44 and @xmath66 we assumed a @xmath43 of 7 k for tmc-1c and 5 k for l1544 for all deuterated isotopologues , following gerin et al ( 1987 ) , and 8 k for tmc-1c and 6 k for l1544 for the main species and the @xmath9c isotopologues as they trace also warmer regions of the cloud .
the effect of the excitation temperature on the derived column densities ratios in table 2 was found to be small , with a change of few percent upon a variation of @xmath671 k. by using these expressions we assumed that the source fills the beam , and optically thin emission obeying lte . since lines of c - c@xmath0h@xmath1 are optically thick , we derived its total column density from the total column density of c - h@xmath9cc@xmath1h assuming a @xmath68c/@xmath9c ratio of 77 , determined by wilson & rood ( 1994 ) from h@xmath1co and co as a function of distance from the galactic center , @xmath69 and @xmath66 were calculated backwards . ccccccccccccccc molecule & transition & frequency & ref.@xmath70 & e@xmath71 & t@xmath72&rms & w&b@xmath73&@xmath74 & v@xmath75&@xmath76 & n@xmath77 & n@xmath78@xmath79 & @xmath80 + & ( ortho / para)&(ghz ) & & ( cm@xmath22)&(k)&mk & ( k km s@xmath22)&%&(arcsec)&(km s@xmath22 ) & ( km s@xmath22 ) & ( @xmath82 @xmath83 ) & ( @xmath84 @xmath83 ) & + * tmc- 1c + c - c@xmath0h@xmath1 & 2@xmath36 - 1@xmath85 ( o ) & 85.338 & 1 & 4.48 & 2.91&7&1.05(1)&81&29 & 5.996(2 ) & 0.338(4 ) & 25(1 ) & 22(1 ) & 1.887 + & 2@xmath36 - 1@xmath85 ( o ) & 85.338 & 1 & 4.48 & 1.27 & 7&0.414(9)&81&29&5.361(4 ) & 0.307(8 ) & & & + & 3@xmath86 - 3@xmath37 ( p ) & 84.727 & 1 & 11.21 & 0.16&7&0.047(2)&81&29 & 5.984(6 ) & 0.27(2 ) & 3.5(2 ) & 22(1 ) & 0.148 + c- h@xmath9cc@xmath1h & 2@xmath36 - 1@xmath85 & 84.185 & 2 & 4.40 & 0.22 & 7&0.060(3)&81&29 & 5.977(7 ) & 0.26(1 ) & 0.52(2 ) & 0.62(3 ) & 0.056 + c - c@xmath0hd & 2@xmath87 - 1@xmath33 & 95.994 & 2&5.25 & 0.1&7 & 0.033(2)&80&27 & 6.03(1 ) & 0.33(3 ) & 1.8(1 ) & 2.8(2 ) & 0.026 + & 3@xmath35 - 2@xmath36 & 104.187 & 2 & 7.54 & 0.34&7&0.091(2 ) & 79&25 & 6.034(3 ) & 0.250(6 ) & 0.63(1 ) & 1.1(3 ) & 0.092 + c - c@xmath0d@xmath1 & 3@xmath35 - 2@xmath36 ( p ) & 94.371 & 3 & 6.84 & 0.04 & 2&0.009(1)&80&27 & 6.07(1 ) & 0.23(2)&0.063(5 ) & 0.17(1)&0.010 + & 3@xmath37 - 2@xmath38 ( o ) & 97.761 & 3 & 6.87 & 0.07 & 2&0.017(1)&80&26 & 6.062(5 ) & 0.233(9 ) & 0.11(4 ) & 0.15(6)&0.018 + & 2@xmath88 - 1@xmath33 ( p ) & 108.654 & 3 & 5.49 & 0.02&2&0.006(1)&78&24&6.06(1 ) & 0.25(3 ) & 0.032(4 ) & 0.09(1 ) & 0.005 + * l1544 + c - c@xmath0h@xmath1 & 2@xmath36 - 1@xmath85 ( o ) & 85.338 & 1 & 4.48 & 2.44&10&1.35(1 ) & 81&29&7.180(2 ) & 0.520(4 ) & 50(2 ) & 37(1 ) & 3.579 + & 3@xmath86 - 3@xmath37 ( p ) & 84.727 & 1 & 11.21 & 0.21&10&0.10(1)&81&29 & 7.210(8 ) & 0.46(1 ) & 4.7(2 ) & 37(1 ) & 0.172 + c- h@xmath9cc@xmath1h & 2@xmath36 - 1@xmath85 & 84.185 & 2 & 4.40 & 0.19 & 10&0.093(3)&81&29 & 7.154(8)&0.44(2 ) & 0.92(4 ) & 0.96(4 ) & 0.096 + c - c@xmath0hd&2@xmath87 - 1@xmath33 & 95.994&2 & 5.25 & 0.13 & 10&0.065(3)&80&27 & 7.17(1 ) & 0.48(3 ) &
4.1(2 ) & 6.2(3 ) & 0.066 + & 3@xmath35 - 2@xmath36 & 104.187 & 2 & 7.54 & 0.48&10&0.238(4 ) & 79&25&7.181(4 ) & 0.468(9 ) & 2.1(4 ) & 4.5(9 ) & 0.278 + c - c@xmath0d@xmath1 & 3@xmath35 - 2@xmath36 ( p ) & 94.371 & 3 & 6.84 & 0.07&5&0.032(2)&80&27&7.20(1 ) & 0.45(3 ) & 0.26(2 ) & 0.77(5 ) & 0.035 + & 3@xmath37 - 2@xmath38 ( o ) & 97.761 & 3 & 6.87 & 0.13&5&0.059(2)&80&26&7.181(7 ) & 0.43(2 ) & 0.44(2)&0.66(2 ) & 0.067 + & 2@xmath88 - 1@xmath33 ( p ) & 108.654 & 3 & 5.49 & 0.04 & 5&0.023(2)&78&24&7.17(2 ) & 0.54(5 ) & 0.16(1 ) & 0.45(4 ) & 0.020 + * * c|cc & * tmc-1c&*l1544 + & ( 0.4 - 0.8)%&(1.2 - 2.1)% + /&(3 - 15)%&(7- 17)% + @xmath89&(5 - 13)%&(12 - 17)% + @xmath89 & - & 4%@xmath90 + @xmath89 & 2%@xmath91&4%@xmath92 + @xmath89 & 8%@xmath91 & 20%@xmath92 + @xmath89 & 1%@xmath91 & 13%@xmath70 + * * d@xmath1 . the color scale gives the 3@xmath5 - 2@xmath6/2@xmath39 - 1@xmath40 line ratio , while the observed value is drawn in as a red contour .
the observed integrated line intensities of 3@xmath5 - 2@xmath6 and 2@xmath39 - 1@xmath40 are shown as blue and green contours , respectively , and the calculated @xmath93 values as grey scale . ] | we report the first interstellar detection of c - c@xmath0d@xmath1 .
the doubly deuterated cyclopropenylidene , a carbene , has been detected toward the starless cores tmc-1c and l1544 using the iram 30 m telescope .
the @xmath2 , @xmath3 , and @xmath4 transitions of this species have been observed at 3 mm in both sources .
the expected 1:2 intensity ratio has been found in the 3@xmath5 - 2@xmath6 and 3@xmath7 - 2@xmath8 lines , belonging to the para and ortho species respectively .
we also observed lines of the main species , c - c@xmath0h@xmath1 , the singly deuterated c - c@xmath0hd , and the species with one @xmath9c off of the principal axis of the molecule , c - h@xmath9cc@xmath1h .
the lines of c - c@xmath0d@xmath1 have been observed with high signal to noise ratio , better than 7.5@xmath10 in tmc-1c and 9@xmath10 in l1544 .
the abundance of doubly deuterated cyclopropenylidene with respect to the normal species is found to be ( 0.4 - 0.8)% in tmc-1c and ( 1.2 - 2.1)% in l1544 .
the deuteration of this small hydrocarbon ring is analysed with a comprehensive gas - grain model , the first including doubly deuterated species .
the observed abundances of c - c@xmath0d@xmath1 can be explained solely by gas - phase processes , supporting the idea that c - c@xmath0h@xmath1 is a good indicator of gas - phase deuteration . |
this paper is part of an ongoing series of papers studying the general h i and optical properties of late - type barred spiral galaxies ( sbc - sd , hereafter ltbs ) .
we will examine the stellar and star forming properties of ngc 925 based on observations from the wiyn 3.5 m telescope .
we previously observed ngc 925 and ngc 1744 , both ltbs , in h i as part of a study of the gaseous properties of this class of galaxies ( pisano _ et al .
_ , 1998 , hereafter paper i ) , and to determine the pattern speed of their bars ( elmegreen _ et al .
ngc 925 was chosen for study because it is a nearby ( @xmath19.3 mpc , silbermann _ et al .
_ 1996 ) , apparently prototypical ltbs which is well oriented in inclination for dynamical and morphological study .
this value for the distance will be used throughout this paper . in paper
i , we found that ngc 925 had a weak spiral pattern and bar as indicated by the small streaming motions in the spiral arms and bar .
asymmetries are quite prevalent in ngc 925 .
ngc 925 s southern spiral arm has reasonably coherent streaming motions , while the northern arms do not .
the center of the bar is also slightly offset from the dynamical center of the galaxy by @xmath1950 pc .
these asymmetries could possibly be related to the presence of a small h i cloud ( m@xmath210@xmath3m@xmath4 ) interacting with the main galaxy .
the strength of the interaction , however , is most likely not enough to drive the asymmetries in ngc 925 .
the presence of an off - center bar , and a single coherent spiral arm are typically viewed as characteristic properties of barred magellanic ( sbm ) galaxies ( de vaucouleurs & freeman 1972 ) , and have been shown to be potentially long - lasting ( levine & sparke 1998 , noordermeer _ et al .
_ 2000 ) while ngc 925 does not illustrate these properties as dramatically as the lmc , for example , it certainly indicates that these characteristic asymmetries do not suddenly appear in sbm s , but are rather a continuous change in morphology throughout late - type disk galaxies .
a summary of basic properties of ngc 925 is given in table 1 .
ltbs are an understudied and , hence , a poorly understood class of galaxies .
while bars may play an important role in galaxy evolution by generating gas inflow ( e.g. , matsuda & nelson 1977 , athanassoula 1992 ) , driving spiral structure ( e.g. , sanders & huntley 1976 ; james & sellwood 1978 ) , and flattening the radial abundance gradient ( martin & roy 1994 ) , these effects are shown to exist primarily in early - type barred spirals ( s0-sb ) .
ltbs do not appear to have strong enough bars to cause such effects ( see paper i ) .
previous studies of ltbs have found that late - type bars tend to have exponential light profiles , as opposed to flat profiles , and end well within corotation , as opposed to just inside it ( elmegreen & elmegreen 1985 , elmegreen _ et al .
_ 1998 ) . furthermore , late - type bars are not likely to be driving the spiral pattern ( sellwood & sparke 1988 ) , in contrast to early - type bars ( elmegreen & elmegreen 1989 ) .
star formation properties are also different , with late - type bars tending to be gas - rich and have star formation throughout the bar , while early - type bars have star formation only near the ends of the bar ( phillips 1995 ) .
our observations in paper i of ngc 925 provided no evidence to contradict any of these properties . in this paper
, we will investigate the optical properties of ngc 925 , with a specific eye towards the nature of star formation in the galaxy .
we will see if there are other signatures of asymmetry in ngc 925 that are typically associated with sbm s .
finally , we will try to devine what role the bar plays in determining the properties of the galaxy as a whole . in section 2 of this paper
we discuss the observations and reduction . in section 3
, we discuss the optical properties of galaxy as a whole .
section 4 discusses the star formation properties of ngc 925 both across the entire galaxy , and in the bar and spiral arms specifically .
we conclude in section 5 .
we observed ngc 925 with the wiyn 3.5 m telescope during the nights of october 6 & 7 , 1996 .
the conditions were superb ; all three nights had subarcsecond seeing and were photometric .
we took the images through the harris b , v , and r broadband filters and on and off band h@xmath0 filters using the wiyn imager , a thinned 2048@xmath5 stis ccd with 21@xmath6 m pixels .
the pixels correspond to @xmath7 on the sky .
the gain was 2.8 e@xmath8 per adu , and the read noise was 8 e@xmath8 .
the imager has a total field of view of 6.5@xmath9 .
since ngc 925 has a major axis of approximately 10@xmath9 , we had to mosaic two images to get full coverage of the galaxy .
we observed the east and west sides of the galaxy separately with overlap in the bar region .
the parameters of the observations are listed in table 2 .
we reduced the data in the usual manner .
we took bias images during the afternoon before each night .
we visually inspected the biases , and statistically compared them and found them to be identical for all three nights . this being the case , we combined all 44 bias frames into one image for bias subtraction to maximize the signal to noise .
each afternoon , we took @xmath15 well exposed ( 10000 - 35000 counts ) dome - flats per filter .
we discarded any flats which were significantly deviant from the average .
we averaged the remaining flats together to correct for the response of the ccd .
we switched filter wheels during each night s observations which sometimes caused the dust features to change in location and/or intensity . as some of the flats changed between nights , we only used the flats from the same night to flatten the program images . aside from the off - line h@xmath0 images ,
all images were flat to better than 5% .
the off - line h@xmath0 images were flat to 10% .
the poor quality of the flat - fielding for the h@xmath0-off filter may require sky - flats to improve .
to flux calibrate our images we observed broadband standards in the pg0231 + 051 and sa92 fields from landolt ( 1992 ) , and g191b2b , a spectrophotometric standard from massey _
the photometric solutions for the broadband observations were found using the _ apphot _ package in iraf .
they are as follows : @xmath10 @xmath11 @xmath12 where ob is the instrumental magnitude and x is the airmass .
we then applied these solutions to the bias subtracted , flat - fielded , sky subtracted images using iraf to produce surface brightness maps .
the errors on these equations are roughly 0.05 mag . as this is comparable to our flat - fielding errors , we combine the errors to obtain our total error on the broadband magnitudes , which is approximately 0.07 mag .
sky subtraction was difficult given the small amount of sky present in our images .
there was typically only a small corner of the ccd frame which appeared to contain mostly sky emission .
for each side of the galaxy we used _
imstat _ to determine a mean sky level , and subtracted it from the image before mosaicing .
this value is quite uncertain given the lack of sky present in the images .
because we were not able to reliably determine the sky values in our images , we did not attempt to determine total magnitudes for ngc 925 . the photometric solutions for the line and continuum images were calculated accounting for the airmass of the observations and the response of the line and continuum filters as determined for our observations of the spectrophotometric standard . using the respective exposure times and the fractional contribution of the two [ nii ] lines to the h@xmath0 image
, we then can get the continuum subtracted h@xmath0 image .
based on the spectra in martin & roy ( 1994 ) , we assumed the h@xmath0 line accounted for 81% of the emission in the line filter , with the two [ nii ] lines accounting for 14% and 5% of the emission .
we applied these solutions to each filter and each side of the galaxy separately .
the line and continuum images were aligned to better than 1 pixel with similar , but not exactly the same , seeing conditions . once each image was fully calibrated , we aligned and averaged together the east and west images of ngc 925 to form a total image of the galaxy . to align the images , we identified stars present in both images , calculated the average shift needed to bring the two sides into alignment , and shifted the images using _
imshift_. images were aligned to better than 1 pixel in each dimension .
a similar procedure was followed to align the full galaxy b , v , r , and h@xmath0 images to each other . at this point
we began our analysis .
figure 1 shows our combined b , v , r image of ngc 925 .
one of the key thrusts of investigations of late - type spirals is the degree to which their disks are lopsided .
extreme late - types such as the magellanic spirals have dynamical centers that are significantly offset from the center of the outer isophotes ( de vaucouleurs & freeman 1972 , odewahn 1996 ) .
we can measure a photometric center from our mosaiced image of ngc 925 and a dynamical center from the hi velocity field ( paper 1 ) .
ngc 925 is also a barred galaxy , so we can compare the photometric and dynamical center with the apparent center of the bar .
magellanic spirals in particular host bars that are greatly offset from the photometric _ and _ dynamical centers .
we chose to use the isophote fitting routine in the iraf package , _
ellipse_. we first removed the stars and saturated columns from the image by replacing the central regions of bright stars within a radius of 2.4@xmath13 with the mean from an annulus with inner radius of 3@xmath13 and width 1@xmath13 . while some stars were still visible in the image the residuals were only slightly higher than the background level of the galaxy .
we fit a series of elliptical annuli 0.@xmath134 in width to the galaxy .
the semi - major axis of each annulus was 4@xmath13 larger than the previous one .
the task ceased to produce fits which converged at a radius of 5@xmath14 , near the edge of our field of view .
we present the results of our elliptical fits in figures 2 & 3 .
note that the scale on the surface brightness plot comes from our photometric calibrations of the background sky level ( although this may not be the true sky ) .
a number of features stand out in the isophotal fits in figure 2 .
overall , the results of the fits are very similar in all bands even though the center , position angle , and ellipticity of the galaxy all vary widely as a function of radius . within a radius of about 60@xmath13 ( i.e. the bar region ) the center
is fit consistently in all bands . simply taking the means from each band we define the bar center to be : 2@[email protected]@xmath170.2@xmath18 , 33@xmath1921@xmath1415@xmath201@xmath13 .
the epoch for all of the coordinates in this paper is 1950 , unless otherwise stated .
there is a sharp change in the fitted declination of the center just beyond a radius of 60@xmath13 , coincident with the position of the prominent spiral arm on the southern side of the galaxy .
the width of the `` bump '' in the fitted declination of the center is consistent with the width of the arm .
the fitted declination then returns to a value comparable with that derived from the innermost parts of the galaxy .
so if we were to ignore the effect of the spiral arm we find that the fitted photometric center of ngc 925 is only slowly varying with radius out to @xmath21 which is close to edge of the optical disk . beyond about 200@xmath13
the center , position angle , and ellipticity remain relatively constant .
while the three bands return inconsistent fits for the center of the galaxy , we take the mean center derived from fits to the annulus 210@xmath13 - 270@xmath13 to be the center of the outer isophotes .
this yields an isophotal center at 2@[email protected]@xmath170.6@xmath18 , 33@xmath1921@[email protected]@xmath13 which is different from the bar center .
studies of lopsided galaxies usually use the center of the outermost isophotes to define a photometric center of the galaxy ( e.g. odewahn 1991 ) .
ngc 925 shows that this center will be a function of the band with which one chooses to carry out the photometry .
the fitted position angle of ngc 925 in figure 2c slowly increases from -80@xmath19 to -55@xmath19 from the center to a radius of 180@xmath13 ; beyond that it drops abruptly .
similar trends are seen in figure 2d where we plot the fitted ellipticity as a function of radius .
it is slowly decreasing from the center out to 180@xmath13 where it also drops abruptly before leveling off .
we list our adopted values for the center , position angle , and ellipticity in table 3 in figure 3a we plot the azimuthally averaged surface brightness distribution for ngc 925 . we fit a double exponential to this distribution and we show the residuals in figures 3b and 3c , respectively .
residuals from our fit solely to the exponential outer - disk ( fig 3b ) clearly illustrate the presence of a second exponential at radii within 60@xmath22 , and a systematic surface brightness enhancement between 140@xmath22 and 180@xmath22 .
the latter is much clearer after we subtract our exponential fit to the bar and re - scale the residuals in the lower panel .
we chose to fit an exponential to the bar , both because it represents the data well , and because elmegreen & elmegreen ( 1985 ) found that late - type galaxies including ngc 925 have exponential bars .
both of our fits are shown in the top panel .
the derived scale - lengths and central brightnesses for the bar and disk are listed in table 3 .
one very interesting property illustrated by these scale - lengths is that the disk tends to get redder with increasing radius , as the b scale - length is much smaller than the r scale - length .
this is contrary to what is found for most field galaxies ( vennik _ et al .
_ 1996 ) , and could be indicative of a differing star formation history for ngc 925 as compared to most other galaxies .
this could also indicate the presence of a very young inner disk and bar in ngc 925 . by looking at all the results of our isophotal fitting
, we can pick out some distinct features in ngc 925 which are evident in both a contour plot of the model ( figure 4 ) , and in the graphs of the fit parameters ( figures 2 & 3 ) .
the inner bar region , within 30@xmath22 , is characterized by widely varying values for the center , position angle , and ellipticity that are most likely due to patchy star formation and dust in the center of the bar .
we get consistent results for the rest of the bar .
the southern arm strongly affects the fits in the remainder of the disk .
this is evident from the drastic change in center , position angle , and ellipticity at 180@xmath22 where the spiral arm fades ( as shown in the surface brightness fits ) .
the slow increase in position angle with increasing radius illustrates the change from the bar dominated region of the galaxy to the region where the spiral arm dominates .
other variations in the fits and surface brightness residuals can be explained by the small scale variations in brightness caused by h ii regions and patchy dust within ngc 925 .
the bar of ngc 925 , based on our fits shown in figure 3 , extends out to approximately 60@xmath22 ( 2.7 kpc ) .
it is at this point where the exponential profile of the bar begins to meet the exponential disk .
the bar is centered at 2@xmath1524@xmath1617@xmath18 in right ascension , and 33@xmath1921@xmath2315@xmath22 in declination , which is not coincident with the center of the outer isophotes .
this is the mean from the isophotal fits of the center within the inner 60@xmath22 .
the bar has an ellipticity of @xmath10.69 ( b / a = 0.3 ) , and , with a position angle of -75@xmath19 , it is aligned within a few degrees of the major axis of the galaxy .
the alignment of the bar position angle with the galaxy position angle , and the origination of the spiral arms from the end of the bar may indicate that the bar has had some effect on the disk of the galaxy .
we now have four centers derived for ngc 925 : the peak of the brightness distribution ( from the rc3 ) , the dynamical center ( from paper i ) , the center of the bar , and the center of the `` outer '' isophotes ( from above ) . in figure 5
, we compare the locations of these centers as a measure of the asymmetry in the galaxy .
the size of the boxes are representative of the fitting errors ; the ellipse is the beam size from the h i data used to model the rotation curve .
we find that the center of the brightness distribution is nearly coincident with the center of the bar , while the dynamical and outer isophote centers are coincident with each other .
the bar center is not coincident , however , with either the dynamical or outer isophotal center .
this is indicative of an asymmetric galaxy and is quite similar to what is seen in sbm s such as ngc 4618 ( odewahn 1991 ) which has a 662 pc displacement between the bar center and the center of the outer isophotes . for ngc 925
the dynamical and outer isophotal centers are offset parallel to the bar , whereas most sbm s show bars offset perpendicular to their major axis ( odewahn 1991 ) .
ngc 925 may be exhibiting a different type of phenomenon , but this property of sbm s requires further study . while we were unable to calculate total magnitudes for ngc 925 ,
we used published values for m@xmath24 to examine mass - to - light ratios for ngc 925 . from the rc3
, we find that ngc 925 has an m@xmath25=-19.9 mag .
using the total and h i masses we derived in paper i , we calculated m@xmath26/l@xmath25 to be 5.0 , and m@xmath27/l@xmath24 to be 0.36 , consistent with what we expect from its classification as an scd spiral ( roberts & haynes 1994 ) , but also similar to values found for later - type spiral galaxies .
in a seminal paper , kennicutt ( 1989 ) observationally examined the dependence of star formation on the properties of the gas in the galaxy . for his study
he used the expression for the critical density for a instability to grow in a thin isothermal gas disk : @xmath28 where c is the velocity dispersion of the gas , g is the gravitational constant , @xmath0 is a dimensionless constant near unity which accounts for the deviation of the disk from a thin , isothermal gas disk , and @xmath29 is the epicyclic frequency given by : @xmath30 where v is the rotation velocity at a radius r. in our case the rotation curve is given by a brandt rotation curve derived from a fit to the h i data in paper i : @xmath31^\frac{3}{2n}}\ ] ] where v@xmath32=118 km s@xmath33 , r@xmath32=17.8 kpc , and n=1.46 ( paper i ) .
kennicutt examined how star formation compared to the gas density and the expected critical density within a collection of galaxies .
he found that star formation , as traced by h@xmath0 emission , was correlated with the gas surface density ( as traced by h i emission ) and that star formation was suppressed when the gas surface density fell below a critical value .
based on data from paper i , and using our h@xmath0 images , we carried out a similar analysis on ngc 925 . from the h i data we derive an average velocity dispersion ,
c , of 10 km s@xmath33 for ngc 925 . the value of @xmath0 should be close to one , because galaxies should be close to being thin , isothermal disks , although kennicutt ( 1989 ) finds an @xmath0 of 0.67 fits his data better . in figure 6
we show the critical surface density ( @xmath34 ) for three values of @xmath0 and surface density of h i multiplied by 1.47 , @xmath35 , ( following the example of kennicutt 1989 to account for the molecular component of the gas ) .
we also plot the h@xmath0 emission normalized to its peak .
the h@xmath0 and h i surface densities shownare azimuthal averages .
we find the h@xmath0 surface brightness to be relatively constant throughout the galaxy before cutting off abruptly at about 15 kpc , whereas the gas surface densitydecreases slowly with radius out to 18 kpc .
this suggests that star formation in ngc 925 is not directly correlated with the gas surface density on a global scale .
while the abrupt decline of h@xmath0 surface brightness occurs at the edge of our field of view , we can use the value of @xmath35/@xmath34 at this point to yield @xmath0 .
looking figure 6 , we see that a value of @xmath0 of near 0.3 works best for ngc 925 as opposed to @xmath0=1 and @xmath0=0.67 . in reality
, the h@xmath0 emission may continue beyond the edge of our field of view , implying a lower value of @xmath0 still .
regardless of the extent of h@xmath0 emission beyond the region we imaged , the star formation criterion found by kennicutt ( 1989 ) , @xmath0=0.67 , does not hold for ngc 925 ; instead , a value of @xmath0 closer to 0.3 is more appropriate .
this has been found to be the case for other galaxies as well .
values of @xmath0 around 0.3 have been found for dirr galaxiesby hunter , elmegreen , & baker ( 1998 ) , and for low surface brightness dwarf galaxies by van zee _
this is evidence that star formation in ngc 925 is governed more by local conditions than by global instabilities ; a property more typical of later type galaxies than ngc 925 .
this is not surprising as star formation can occur in any type galaxy even when such a criterion is not met ( see ferguson _
_ 1998 for a discussion ) .
_ find h ii regions outside the cited limits to the optical disk in a sample of spiral galaxies .
their outer h ii regions , and ours , lie on coherent spiral arms even at large radii .
a better analysis of star formation criteria in ngc 925 could be done using co data to trace the molecular component of the gas and using a larger field - of - view ccd to better trace the full extent of the h@xmath0 emission in ngc 925 .
comparing the h i distribution of paper i with the h@xmath0 and broadband images of ngc 925 allows us to examine the nature of star formation in the galaxy .
the star formation in ngc 925 , as traced by the h@xmath0 emission , is found in two main regions .
the largest star forming regions are located at the end of the southern spiral arm and in the bar . in addition
, star formation occurs at a lower level throughout the southern spiral arm ( although it may be partially obscured by dust ) .
the northern arm has sporadic star formation near the bar , but very little elsewhere , except , perhaps , near the end of the arm on the east side of the galaxy .
the lack of coherent star formation along the north arms is not surprising given the weak definition of the stellar and gaseous arms .
figure 7 shows the h@xmath0 in red on the h i in blue .
in general , the star formation is occurring on top of h i peaks and the h@xmath0 emission traces the h i spiral arms very closely .
the coincidence between the two distributions is highly correlated across the entire galaxy , but the exceptions are quite interesting .
specifically , there is a sizable h ii region on the northwest spiral arm near the bar which is sitting in a local h i depression .
this is the only major h ii region which is not associated with a h i peak .
this could be due to the star formation depleting the h i in this region , ionizing it , or the h i hole could be filled with molecular gas .
looking at the broadband bvr image with the h i contours ( figure 8) , we see that the peak h i does trace the optical emission ( as well as the h@xmath0 emission ) , as expected .
this figure clearly shows that the southern spiral arm is much stronger and more coherent than the northern arm in h i and optical emission .
the north side of the galaxy shows more flocculent spiral structure and less intense star formation as compared to the south .
the southern arm also has a prominent gap in optical emission in the middle of the arm , while it still has a large amount of h i. this may be due to a large dust lane crossing the arm at this position .
these are not caused by inclination , as the southern part of the galaxy is the far side of the galaxy .
you can also see the bluest regions of the galaxy are those where large h ii regions exist , specifically at the end of the southern arm , and to the northeast of the bar ( which is the end of one of the northern arms ) .
as our data is of sufficient resolution and photometric quality to address the question of triggering of star formation in spiral arms of late - type galaxies , we took cuts through various regions of the spiral arms and bar in ngc 925 ( figure 10 ) to look for offsets between the peaks of emission in b , r , h@xmath0 , and h i ( see beckman & cepa 1990 ) .
figure 9 shows the location of these cuts .
these cuts were 4@xmath22 wide and of varying length ( as shown in figure 9 ) .
they are roughly centered on the arms or bar .
the emission is an average over the width of the cut and is normalized to the peak intensity of that band in the cuts . triggering is defined as an increase in the star formation efficiency in the arm versus the inter - arm region ; it is not simply an increase in total star formation .
an offset in the peaks would suggest there is an age gradient across the cut ( in a dust free world ) .
beckman & cepa ( 1990 ) , for instance , see such a color gradient in ngc 7479 between the b and i bands .
dust , however , can produce a similar gradient due simply to an extinction gradient within the spiral arm ( elmegreen 1995 ) .
in addition , we can not determine if triggering is actually occurring in the arm solely from an age ( or color ) gradient ; it is entirely possible to see a gradient without any triggering ( cf .
elmegreen 1995 ) .
what we can say is that the absence of an offset between the peaks in b and r implies the lack of an age gradient in the underlying stellar population .
cuts along the southern arm ( figure 10a - d ) show the optical emission is well - aligned with the h i across the arm for most of the length of the arm .
peaks in b and r are well - aligned . in general , the h@xmath0 peaks correlate with the broadband peaks , but not always with the h i peaks .
there are regions with no h@xmath0 peaks , but with strong broadband emission .
in addition , there are two noticeable instances of h@xmath0 peaks in h i depressions .
these occur at @xmath36 40@xmath37 in cut c and in cut d at 40@xmath37 .
star formation in the south arm is located in large h ii regions which tend to lie on the peak of the h i distribution .
cuts across the north arms ( figure 10e - g ) of ngc 925 show a decidedly different phenomenon than those of the south arm .
cut e shows optical emission peaking in an h i depression at -50@xmath22 .
there is also some h@xmath0 emission peaking with the h i at 20@xmath22 .
other cuts across the north arms , such as cut f , show a correlation between the optical and gas peaks , while cut g shows an offset between the optical emission and the h i peaks . in the north arms
, the star formation appears less prolific than in the southern arm and occurs in smaller h ii regions which are less coherently placed than in the south of ngc 925 .
this suggests that the nature of the spiral pattern , and hence the star formation , in the north and south arms of ngc 925 are different , if not fundamentally , then in our perception of it .
this type of asymmetry , with one spiral arm being dominant in a galaxy , is quite typical in later - type spiral galaxies , such as sbm s ( odewahn 1991 , de vaucouleurs & freeman 1972 ) .
ngc 925 illustrates that these properties are not limited to sbm s , but are evident in earlier type spiral galaxies , although less prominently
. in general , ltbs tend to have gas - rich bars with star formation occurring throughout the bar ( phillips 1995 ) .
friedli & benz ( 1995 ) suggest that star formation occurs along the major axis of dynamically young bars , and is concentrated in the center or in rings of older bars .
ngc 925 s bar is quite gas - rich and has star formation occurring all along its major axis .
thus , the friedli & benz ( 1995 ) model would imply that ngc 925 has a young bar .
the extremely blue center of ngc 925 , compared with its outer region ( see section 3 ) , could indicate a dynamically young bar as well , one which has just recently started prolific star formation . on the other hand , it is possible that weak bars , such as ngc 925 s , are simply inefficient at funneling gas into the galactic nucleus . furthermore , because corotation is far out in the disk ( elmegreen _ et al .
_ 1998 ) , the spiral arms have a large reservoir of gas to drive towards the center of ngc 925 . as there is no inner lindblad resonance in ngc 925 ( elmegreen _ et al .
_ 1998 ) where the gas would pile up , the bar would be well - supplied with gas for long - term star formation , and it may not be young at all .
the star formation in the bar appears to be offset somewhat from the h i peaks .
examining the bar region in figure 7 , we find that there is a great deal of star formation on the north side of the bar that is systematically displaced away from the h i peaks .
this is better illustrated by cuts h - k in figure 10 .
furthermore , the minor axis cuts ( i - k ) show that the b and r peaks are systematically offset to the north of the h i peak .
while there is a h@xmath0 peak aligned with the h i peak in cut j , the main optical emission is offset .
this could be explained by dust extinction , depletion of the h i by star formation or ionization , the gas being in molecular form , or the bar being a wave phenomenon . for this offset to be an extinction feature , some process must pile up the dust preferentially along one side of the bar .
dust lanes have been found in the bars of other galaxies , and their coherence is usually attributed to shocks occurring in these bars ( see athanassoula 1992 for a nice discussion ) .
the bar of ngc 925 , however , does not have strong streaming motions ( paper i ) which would be indicative of shocks .
the dust appears to be in filaments in the bar region , and not in well defined lanes .
furthermore , some of the cuts perpendicular to the major axis of the bar do show star formation correlated with the h i peaks , but all of the cuts indicate that the stellar emission is downstream of the h i peaks .
extinction should affect both h@xmath0 emission and r band emission to a similar extent .
as this is not evident in the cuts , it is improbable that dust is causing the apparent offset .
the major axis cut ( h ) shows the h@xmath0 peaking throughout the bar , with the gaps in emission likely due to the prolific dust content of the bar .
however , the broadband optical peaks are concentrated in a h i depression
. this may be due to the recent star formation and young stars evacuating or ionizing the h i in this region of the bar .
the h i depression in the bar could be filled with molecular gas , and not truly be depleted of all gas . along the major axis of the bar the h
i seems to be piled up at the bar ends , particularly near the dynamical center of the galaxy at 50@xmath22 .
spectra of the bar taken with the densepak instrument on wiyn are in hand and should yield insight into the nature of the ionized gas in the bar .
higher spatial resolution h i data will also help to untangle the dynamics of gas in the bar . if the offset of h i south of the h@xmath0 in the bar is not caused by dust , or by young stars ionizing or otherwise clearing out the h i in the bar , or by the gas being in molecular form , then the bar could be a wave phenomenon .
this may be linked to the offset dynamical center of the galaxy as reported in paper i. the dynamical center is located on the west end of the bar , so the geometry and dynamics make sense for this scenario , with the star formation occurring downstream of the gaseous peaks and the stars even further downstream , provided that star formation is occurring as the gas moves through the wave
we have observed ngc 925 using the wiyn telescope in b , v , r , and h@xmath0 filters to better understand the stellar distribution and star formation properties of the galaxy .
we have compared these observations with previously described h i observations ( paper i ) to get a nearly complete picture of the properties of ngc 925 .
the global properties of ngc 925 are typical for a late - type spiral galaxy .
it has an obvious bar , and two large spiral arms .
the southern spiral arm is apparently much stronger than the northern arm , which is quite flocculent .
the relative dominance of the southern arm is apparent both optically and in the h i ( paper i ) .
this difference between the northern and southern spiral arms is not the only asymmetry present in the galaxy .
the center of the galaxy , as derived from the outer isophotes , is coincident with the dynamical center ( paper i ) , but not with the center of the bar .
these properties are typically associated with barred magellanic spirals ( de vaucouleurs & freeman 1972 ) , although in the case of sbm s the offset bar and single , dominant spiral arm tend to be much more pronounced than they are in the case of ngc 925 .
nevertheless , we see that such properties are not limited to magellanic spirals , but are present in more subtle ways in earlier - type spirals .
the optical properties of ngc 925 , such as mass - to - light ratio and absolute magnitude , are as would be expected for a typical late - type galaxy . from our isophote fitting
, we find that ngc 925 s surface brightness distribution can be characterized by a double exponential ; one for the bar and one for the outer disk .
the bar is quite easily distinguished from the rest of the galaxy in the fits , by having a well - defined center , position angle , and ellipticity differing from the outer disk .
the bar also has a exponential brightness distribution with a scale - length much smaller than the rest of the galaxy .
there is little evidence that the bar has any effect on the structure of the rest of the galaxy .
the isophote fits show quite a bit of structure outside of the bar region .
this is almost certainly due to the bright southern spiral arm , which shows up clearly as a surface brightness enhancement in the fits . at the same radius as this enhancement
, we can see distinct changes in the fits for the center , ellipticity , and position angle .
the fits do not seem to be affected by the weaker northern arm .
this is further evidence for ngc 925 s similarity to the sbm one - armed spiral pattern .
the fits also show that ngc 925 gets redder with increasing radius , atypical of normal spiral galaxies in the field .
this could indicate a recent enhancement of star formation in the inner galaxy . from our analysis of the critical density for massive star formation in ngc 925 , using the technique of kennicutt ( 1989 ) , we find that the galaxy has widespread star formation occurring outside of the radius where it should be suppressed .
ngc 925 has a very flat h@xmath0 surface brightness distribution , with emission extending all the way to the edge of our image at 15 kpc .
kennicutt ( 1989 ) using a sample of 15 sc galaxies found that star formation was suppressed when @xmath0=@xmath35/@xmath34 fell below 0.67 .
for ngc 925 star formation continues past this radius and is more consistent with an @xmath380.3 , like what was found by hunter
_ et al . _
( 1998 ) and van zee _
( 1997 ) for later - type galaxies .
the exact value of @xmath0 is unclear , but probably lower , as the h@xmath0 emission could continue beyond our field of view . as mentioned above , the spiral arms in ngc 925 are quite different .
the north arm is quite weak , lacks coherence , and has only patchy star formation .
the southern arm , on the other hand , can be traced all the way from the bar out to the edge of the galaxy and has plenty of star formation . aside from a small gap ,
the arm is quite coherent .
there is a long dust lane associated with the southern arm , which probably obscures some star formation .
if there is dust along the northern arm , it is not nearly so well - behaved .
taking cuts across the northern and southern arms , we can measure the relative positions of the stars , h ii regions , and h i. in the southern arm , the strongest h@xmath0 and stellar emission is closely aligned with the peak of the h i. large scale coherence breaks down in the northern half of the disk . at some places ( 40 - 50@xmath13 in cut f )
star formation is well correlated with peaks in the h i distribution . at other locations ( @xmath140@xmath13 in cut g )
star formation is displaced off of the h i peaks while elsewhere the h i and h@xmath0 are anti - correlated ( e.g. @xmath150@xmath13 in cut e ) .
the difference between the north and south arms suggests that there may be some fundamental difference in the nature of the two arms . using the same cuts
, we looked for evidence of triggering by the spiral arms .
if there is triggering in the arms , we might expect a color gradient ( implying an underlying age gradient ) perpendicular to the arms .
such a color gradient can also be caused by dust extinction .
the lack of a color gradient would imply the lack of an age gradient , and , hence a lack of triggering . while the cuts of the north arms show that the stellar emission is typically offset from the h@xmath0 and h i emission
, there is no offset present between the b & r peaks .
similarly , the southern arm shows no offset between the broadband emission peaks .
the lack of an offset shows that there is no color gradient , and hence it is unlikely that an age gradient due to triggering in the spiral arm is occurring .
this suggests that there is no larger scale organization of the star formation in the northern disk a result reminiscent of irregular galaxies .
star formation must be driven by local conditions and processes in the northern half , while larger scale phenomena organize star formation in the bar and southern arm .
finally , we examined the nature of the ngc 925 s bar and the star formation occurring within it .
as is expected for late - type bars ( phillips 1995 ) , ngc 925 has star formation occurring all along the bar .
cuts perpendicular to the major axis of the bar show that the star formation is occurring along the major axis , but the broadband optical emission is systematically offset to the north of the h i peaks . while this could be a dust effect , we would expect the h@xmath0 emission to be similarly affected , but it is not .
this could also be due to the massive star formation in the bar ionizing or clearing out the h i in the bar .
the systematic offset may be related to the dynamical center being offset from the bar center , and hence the bar could be a wave pattern moving through the gaseous medium in ngc 925 .
we might expect to see the peak of the stellar emission downstream of the star formation , as is the case here .
it may be that the bar of ngc 925 is not actually a bar , but has more in common with the spiral arms of ngc 925 .
overall , this study has shown that ngc 925 is a prototypical late - type spiral galaxy , with some properties that are characteristic of magellanic spirals , such as an off - center bar and a dominant spiral arm .
these traits are less pronounced in ngc 925 than in an sbm indicating a probable smooth transition of these properties from scd galaxies to sm s .
massive star formation in ngc 925 persists beyond the radius predicted by kennicutt ( 1989 ) for late - type spirals , but behaves more like dirr s ( hunter _ et al .
this result may be exacerbated by wider field imaging of ngc 925 , however , and star formation almost certainly continues at a low level beyond this point ( see ferguson _
_ 1998 for an example ) .
the north and south arms are not only different in brightness , but in coherence and in the distribution of stars across the arms .
the north arm has stars offset from h i peaks , while the south arm has both star formation and stellar emission coincident with the neutral gas peaks .
finally , the bar of ngc 925 appears to be a typical late - type bar being gas - rich and having star formation throughout , but the stars are offset from the h@xmath0 and h i peaks .
this fact coupled with the offset dynamical center of ngc 925 suggests that the bar may be more of a wave phenomenon , similar to a spiral arm .
the authors would like to thank the staff at the wiyn observatory and on kitt peak for helping make our observing run go smoothly , and for their excellent assistance in producing the wonderful data from the wiyn telescope .
the authors also thank the anonymous referee for comments improving the quality of this paper .
this work was partially supported by nsf grant ast 96 - 16907 to e.m.w .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
the results from the isophote fitting of ngc 925 .
counter - clockwise from upper left they are right ascension of center , declination of center , ellipticity , and position angle .
blue represents fits to the b band image , green for v , and red for r. the results of the isophote fits to the surface brightness in b , v , r for ngc 925 .
the colors are the same as figure 2 .
the upper panel shows the data on an arbitrary scale ( points ) and the fits to the bar and disk ( solid lines ) .
the middle panel shows the residuals from just the disk fit .
the lower panel shows the residuals from both the bar & disk fits .
the solid line in the lower 2 panels mark the zero level .
a representation of the model isophotes as fit to the r band image of ngc 925 .
ellipses are plotted for one out of every three annuli .
they illustrate the relative location and orientation of each annulus , but do not reflect the surface brightness distribution .
this plot shows 4 centers of ngc 925 : the center of ngc 925 according to the rc3 ( rc3 ) , the center of the bar ( b ) , the center of the outer isophotes ( iso ) , and the dynamical center ( dyn ) .
the boxes represent the derived errors to the fitted centers .
the ellipse is the fwhm of the beam from the h i observations in paper i. \a ) top panel is a plot of gas surface density ( solid line ) , as measured by h i observations from paper i , and multiplied by 1.47 to account for molecular gas .
the dashed line is the critical gas surface density for collapse an @xmath0=1 .
the dot - dash line is the radial h@xmath0 distribution normalized to its peak .
the vertical dotted line indicates the nearest edge of the chip from the center of the bar of ngc 925 .
the bottom panel indicates the ratio of the measured gas surface density to the critical density . the horizontal dotted line is the level when massive star formation should cease .
the vertical horizontal line is the same as in the top panel .
b ) same as in a , but for @xmath0=0.67 .
c ) same as a , but for @xmath0=0.3 .
this figure shows the h i contours from paper i ( in yellow ) overlaid on the combined b , v , r wiyn image from figure 1 .
hatched contours indicate a h i depression . the h i beam is shown in the lower right .
this is an optical r band image of ngc 925 , with the location of the cuts marked on it .
the rectangular boxes are show the region over which the cut was taken .
the letters correspond to the cuts shown in figure 10 , and also marks the most positive point in the cut .
the results of the cuts across the southern arm , northern arms , and bar of ngc 925 .
each band of observations is normalized to its peak .
the solid line is b band , the dashed line is r , the dot - dash line is h@xmath0 , and the dotted line is h i column density .
cuts a - d are for the southern arm .
cuts e - g are for the northern arms .
cut h is the bar major axis , and cuts i - k are parallel to the bar minor axis .
the most positive points in the cuts are marked by the letters in figure 9 . | we present the results from an optical study of the stellar & star formation properties of ngc 925 using the wiyn 3.5 m telescope .
images in b , v , r , & h@xmath0 reveal a galaxy that is fraught with asymmetries . from isophote
fits we discover that the bar center is not coincident with the center of the outer isophotes nor with the dynamical center ( from pisano _ et al .
_ 1998 ) .
cuts across the spiral arms reveal that the northern arms are distinctly different from the southern arm .
the southern arm not only appears more coherent , but the peaks in stellar and h@xmath0 emission are found to be coincident with those of the h i distribution , while no such consistency is present in the northern disk .
we also examine the gas surface density criterion for massive star formation in ngc 925 , and find that its behavior is more consistent with that for irregular galaxies , than with late - type spirals . in particular , star formation persists beyond the radius at which the gas surface density falls below the predicted critical value for star formation for late - type spirals .
such properties are characteristic of magellanic spirals , but are present at a less dramatic level in ngc 925 , a late - type spiral . |
in the last few decades the advances in the observational cosmology have led the field to its `` golden age . ''
cosmologists are beginning to nail down the basic cosmological parameters , and have started asking questions about the nature of the initial conditions provided by inflation @xcite , which apart from solving the flatness and horizon problem , also gives a mechanism for producing the seed perturbations for structure formation @xcite , and other testable predictions .
the main predictions of a canonical inflation model are : ( i ) spatial flatness of the observable universe , ( ii ) homogeneity and isotropy on large scales of the observable universe , ( iii ) nearly scale invariant and adiabatic primordial density perturbations , and ( iv ) primordial perturbations to be very close to gaussian .
the cosmic microwave background ( cmb ) data from the wilkinson anisotropy probe ( wmap ) @xcite , both temperature @xcite and polarization @xcite anisotropies , have provided hitherto the strongest constraints on these predictions @xcite .
there is no observational evidence against simple inflation models .
non - gaussianity from the simplest inflation models that are based on a slowly rolling scalar field is very small @xcite ; however , a very large class of more general models with , e.g. , multiple scalar fields , features in inflaton potential , non - adiabatic fluctuations , non - canonical kinetic terms , deviations from bunch - davies vacuum , among others ( * ? ? ?
* for a review and references therein ) generates substantially higher amounts of non - gaussianity .
the amplitude of non - gaussianity constrained from the data is often quoted in terms of a non - linearity parameter @xmath0 ( defined in section [ model ] ) .
many efficent methods for evealuating bispectum of cmb temperature anisotropies exist @xcite . so far , the bispectrum tests of non - gaussianity have not detected any significant @xmath0 in temperature fluctuations mapped by cobe @xcite and wmap @xcite .
different models of inflation predict different amounts of @xmath0 , starting from @xmath4 to @xmath5 , above which values have been excluded by the wmap data already . on the other hand ,
some authors have claimed non - gaussian signatures in the wmap temperature data @xcite .
these signatures can not be characterized by @xmath0 and are consistent with non - detection of @xmath0 .
currently the constraints on the @xmath0 come from temperature anisotropy data alone . by also having the polarization information in the cosmic microwave background
, one can improve sensivity to primordial fluctuations @xcite .
although the experiments have alrady started characterizing polarization anisotropies @xcite , the errors are large in comparison to temperature anisotropy .
the upcoming experiments such as planck will characterize polarization anisotropy to high accuracy .
are we ready to use future polarization data for testing gaussianity of primordial fluctuations ?
do we have a fast estimator which allows us to measure @xmath0 from the combined analysis of temperature and polarization data ? in this paper
we extend the fast cubic estimator of @xmath0 from the temperature data @xcite and derive a fast way for measuring primordial non - gaussianity using the cosmic microwave background temperature and polarization maps .
we construct a cubic statistics , a cubic combination of ( appropriately filtered ) temperature and polarization maps , which is specifically sensitive to the primordial perturbations .
this is done by reconstructing a map of primordial perturbations , and using that to define our estimator .
we also show that the inverse of the covariance matrix for the optimal estimator @xcite is the same as the product of inverses we get in the fast estimator .
our estimator takes only @xmath6 operations in comparison to the full bispectrum calculation which takes @xmath7 operations . here
@xmath3 refers to the total number of pixels . for planck @xmath8 , and
so the full bispectrum analysis is not feasible while ours is .
the harmonic coefficients of the cmb anisotropy @xmath9 can be related to the primordial fluctuation as : @xmath10 where @xmath11 and @xmath12 are the harmonic coefficients of the primordial curvature perturbations and the primordial isocurvature perturbations respectively at a given comoving distance @xmath13 ; @xmath14 is the radiation transfer function of either adiabatic or isocurvature perturbations ; x refers to either t or e ; and @xmath15 is the bessel function of order @xmath16 . a beam function @xmath17 and the harmonic coefficient of noise @xmath18
are instrumental and observational effects .
( [ phi_alm ] ) is written for flat background , but can easily be generalized . any non - gaussianity present in the primordial perturbations @xmath19 or @xmath20 , can get transfered to the observed cmb i.e. @xmath21 , via eq .
( [ phi_alm ] ) . due to the smallness of isocurvature contribution over the curvature perturbation
@xcite we will drop the isocurvature contribution from eq .
( [ phi_alm ] ) and further we will use a popular and simple non - gaussianity model @xcite given by @xmath22 where @xmath23 is the linear gaussian part of the perturbations , and @xmath0 is a non - linear coupling constatnt characterizing the amplitude of the non - gausianity .
the bispectrum in this model can be written as : @xmath24 the above form of the bispectrum is specific to the model chosen and so in general the constrains on @xmath0 do not necessarily tell us about non - gaussianity of other models @xcite .
this model , for instance , fails completely when non - gaussianity is localized in a specific range in @xmath25 space , the case that is predicted from inflation models with features in inflaton potential @xcite . even for the simplest inflation models based on a slowly rolling scalar field ,
the bispectrum of _ inflaton perturbations _ yields a non - trivial scale dependence of @xmath0 @xcite , although the amplitude is too small to detect . on the other hand the bispectrum of _ curvature perturbations _
contains the contribution after the horizon exit , which is non - zero even when inflaton perturbations are exactly gaussian .
this contribution actually follows eq .
( [ phi_bispec ] ) @xcite .
curvaton models also yield the bispectrum in the form given by eq .
( [ phi_bispec ] ) @xcite .
reconstruction of primordial perturbations from the cosmological data allows us to be more sensitive to primordial non - gaussianity , which is important because non - gaussianity in the cosmic microwave background data does not necessarily imply the presence of primordial non - gaussianity .
the method of reconstruction from temperature data is described in @xcite and that from the combined analysis of temperature and polarization data is described in @xcite , where we reconstruct the perturbations @xmath26 using an operator @xmath27 .
these operators are given by : @xmath28 where @xmath29 @xmath30 @xmath31 is the power spectrum of the primordial curvature perturbations , and @xmath32 is the noise covariance matrix .
the primordial perturbation then can be estimated as : @xmath33 figure 1 shows the improvement in reconstruction due to additional information from the cmb polarization .
we can construct a quantity @xmath34 that is analogous to the ksw fast estimator of @xmath0 from temperature data @xcite but generalize it to include polarization data as well .
this quantity has a simple interpretation in terms of the tomographic reconstruction of the primoridal potential described in @xcite .
it is the radial integral of a cubic combination of the scalar potential reconstructions , the b terms with the analogous a term . as in @xcite , we form a cubic statistics given by @xmath35 where @xmath36 @xmath37 @xmath38 , and @xmath39 is a fraction of sky .
index @xmath40 and @xmath41 can either be @xmath42 or @xmath43 .
we find ( see appendix a ) that @xmath44 reduces to @xmath45 where @xmath46 is : 1 when @xmath47 , 6 when @xmath48 , and 2 otherwise ; @xmath49 is the theoretical bispectrum for @xmath50 , and is given by : @xmath51 where @xmath52 since @xmath53 is proportional to @xmath0 , an unbiased estimator of @xmath0 can be written as : @xmath54 the most time consuming part of the calculation is the harmonic transformation necessary for eqs .
( [ b ] ) and([a ] ) .
the fast estimator defined above takes only @xmath6 operations times the number of sampling points for @xmath55 , which is of the order 100 .
hence this is much faster than the full bispectrum analysis , which scales as @xmath7 .
@xmath3 here is the number of pixels . in the next section
we will show that this fast unbiased estimator is also optimal , by proving equivalence with a known optimal but slow estimator
. in their recent paper babich and zaldarriaga @xcite have found an optimal estimator for @xmath0 which minimizes the expected @xmath56 given by @xmath57 this optimal estimator is @xmath58 where the index @xmath59 and @xmath60 runs over @xmath61 , unlike in the fast estimator case where @xmath59 and @xmath60 run over all the 8 combinations @xmath62 . in appendix
b we show that the inverse of the covariance matrix for the bz estimator is the same as the product of inverses we get in the fast estimator , eq . [ fastestim ] ; hence our estimator is optimal
to test optimality of our fast estimator we use eqs .
( [ fnl_estimate ] ) and ( [ s_prim ] ) to measure @xmath0 from simulated gaussian skies .
the error bars on @xmath0 are then derived from monte carlo simulations of our fast estimator .
the simulated errors are compared with the cramer rao bound @xmath63 , where @xmath39 is a fraction of sky , and @xmath64 is the fisher matrix given by @xcite : @xmath65 where @xmath46 is : 1 when @xmath47 , 6 when @xmath48 , and 2 otherwise .
we have used @xmath66cdm model with @xmath67 , and a constant scalar spectral index @xmath68 . since the contribution to the integral in eqs .
[ b ] and [ a ] comes mostly from the decoupling epoch ( for our simulations @xmath69gpc ) , our integration limits are @xmath70 mpc and @xmath71 mpc , with the sampling @xmath72 mpc .
refining the sampling of the @xmath55 integral by a factor of 2 does not change our results significantly .
the results are summarized in figure [ fig3 ] .
we separately explore the effects of excluding foreground contaminated sky regions and noise degradation due to the instrument .
for definiteness we have used the published planck noise amplitudes which are described in table 2 , though we ignore the effect of the scanning strategy and noise correlations .
we shall come back to this point in [ sec : conclusions ] .
we find that the low @xmath16 modes of the polarization bispectrum are contaminated significantly when @xmath73 .
this is illustrated by the lower right panel of figure [ fig3 ] , where we sum over all the @xmath16 modes for @xmath74 contribution to the bispectrum ( because the temperature bispectrum is not contaminated by the sky cut ) , while varying @xmath75 for the other terms ( @xmath76 ) .
our results show that one may simply remove the contamination in the polarization bispectrum due to the sky cut by removing @xmath77 ( when @xmath78 ) without sacrificing sensitivity to @xmath0 .
the contamination appears to be less when noise is added , as the dominant constraint still comes from the temperature data which are insensitive to the contamination due to the sky cut .
we can explain this behaviour simply in terms of the coupling between spherical harmonic modes induced by the sky cut .
the key observation is that the low @xmath16 polarization spectrum is very small in the theoretical model we chose , but if it could be observed it would add information to the @xmath0 estimator .
therefore , for a noiseless experiment , an optimal estimator designed for full sky work will assign a large weight to the low @xmath16 polarization modes .
since the low @xmath16 spectrum rises very steeply towards higher @xmath16 a sky cut creates power leakage from higher @xmath16 to lower @xmath16 .
this biases the low @xmath16 spectrum significantly and this bias is amplified by the large coefficient the estimator assigns to these modes . in a realistic experiment including noise , the low @xmath16 polarization modes are difficult to measure , since noise dominates .
accordingly , the estimator assigns small weights to the low @xmath16 polarization modes and the sky cut has a much less prominent effect .
our fast estimator takes only @xmath6 operations times the number of sampling points for the integral over @xmath55 , which is of the order 100 .
hence this is much faster than the full bispectrum analysis as discussed in @xcite , which goes as @xmath7 .
@xmath3 here is the number of pixels . for planck
we expect @xmath79 , so performing 100 simulations using 50 cpus takes only 10 hours using our fast estimator , while we estimate it would take approximately @xmath80 years to do the brute force bispectrum calculation using the same platform .
.planck noise properties assumed for our analysis[tbl-2 ] [ cols="^,^,^,^,^,^,^,^ " , ] +
starting with the tomographic reconstruction approach @xcite we have found a fast , feasible , and optimal estimator of @xmath0 , a parameter characterizing the amplitude of primordial non - gaussianity , based on three - point correlations in the temperature and polarization anisotropies of the cosmic microwave background .
using the example of the planck mission our estimator is faster by factors of order @xmath81 than the estimator described by babich and zaldarriaga @xcite , and yet provides essentially identical error bars .
the speed of our estimator allows us to study its statistical properties using monte carlo simulations .
we have explored the effects of instrument noise ( assuming homogeneous noise ) , finite resolution , as well as sky cut .
we conclude that our fast estimator is robust to these effects and extracts information optimally when compared to the cramer rao bound , in the limit of homogeneous noise .
we have uncovered a potential systematic effect that is important for instruments measuring polarization with extremely high signal - to - noise on large scales .
the inevitable removal of contaminated portions of the sky causes any estimator based on the pseudo - bispectrum to be contaminated by mode - to - mode couplings at low @xmath16 .
we have demonstrated that by simply excluding low @xmath16 polarization modes from the analysis removes this systematic error with negligible information loss .
it has been shown that inhomogeneous noise causes cubic estimators based on the psudo - bispectrum with a flat weighting to be significantly suboptimal @xcite .
a partial solution to this problem has been found by @xcite , where a linear piece has been added to the estimator in addition to the cubic piece .
it should be straightforward to apply their method to our estimator .
finally , our reconstruction approach may be extended to find fast estimators for higher order statistics , for example trispectrum based estimators of @xmath0 @xcite .
this is the subject of ongoing work .
we acknowledge stimulating discussions with michele liguori .
ek acknowledges support from an alfred p. sloan fellowship .
bdw acknowledges support from the stephen hawking endowment for cosmological research .
some of the results in this paper have been derived using the cmbfast package by uros seljak and matias zaldarriaga @xcite and the healpix package @xcite .
this work was partially supported by the national center for supercomputing applications under tg - mca04t015 and by university of illinois .
we also utilized the teragrid cluster ( www.teragrid.org ) at ncsa .
bdw and apsy s work is partially supported by nsf ast o5 - 07676 and nasa jpl subcontract 1236748 .
natexlab#1#1 acquaviva , v. , botrolo , n. , matarrese , s. , & riotto , a. 2003 , nucl .
b , 667 , 119 albrecht , a. & steinhardt , p. j. 1982 , phys .
lett . , 48 , 1220 , d. , creminelli , p. , & zaldarriaga , m. 2004 , journal of cosmology and astro - particle physics , 8 , 9 , d. & zaldarriaga , m. 2004 , , 70 , 083005 bardeen , j. m. , steinhardt , p. j. , & turner , m. s. 1983 , phys .
d. , 28 , 679 , n. , komatsu , e. , matarrese , s. , & riotto , a. 2004 , , 402 , 103 , r. , dunkley , j. , & pierpaoli , e. 2006 , , 74 , 063503 , c. l. , halpern , m. , hinshaw , g. , jarosik , n. , kogut , a. , limon , m. , meyer , s. s. , page , l. , spergel , d. n. , tucker , g. s. , wollack , e. , wright , e. l. , barnes , c. , greason , m. r. , hill , r. s. , komatsu , e. , nolta , m. r. , odegard , n. , peiris , h. v. , verde , l. , & weiland , j. l. 2003 , , 148 , 1 , p. , hansen , f. k. , liguori , m. , marinucci , d. , matarrese , s. , moscardini , l. , & vittorio , n. 2006 , , 369 , 819 . 2006 , , 369 , 819 , g. & szapudi , i. 2006 , , 647 , l87 , x. , easther , r. , & lim , e. a. 2006 , arxiv astrophysics e - prints , astro - ph/0611645 chiang , l. y. , naselsky , p. d. , & coles , p. 2004
, astrophys .
j , 602 , 1 chiang , l. y. , p. d , n. , verkhodanov , o. v. , & way , m. j. 2003 , astrophys .
j , 590 , l65 , p. , nicolis , a. , senatore , l. , tegmark , m. , & zaldarriaga , m. 2006 , journal of cosmology and astro - particle physics , 5 , 4 , p. , senatore , l. , zaldarriaga , m. , & tegmark , m. 2006 , arxiv astrophysics e - prints , astro - ph/0610600 falk , t. , rangarajan , r. , & srednicki , m. 1993 , astrophys .
j. , 403 , l1 gangui , a. , lucchin , f. , matarrese , s. , & mollerach , s. 1994 , astrophys . j. , 430 , 447 , k. m. , hivon , e. , banday , a. j. , wandelt , b. d. , hansen , f. k. , reinecke , m. , & bartelmann , m. 2005 , , 622 , 759 guth , a. h. 1981 , phys .
d , 23 , 347 guth , a. h. & pi , s. y. 1982 , phys .
lett . , 49 , 1110 hawking , s. w. 1982 , phys .
b. , 115 , 295 , g. , nolta , m. r. , bennett , c. l. , bean , r. , dore , o. , greason , m. r. , halpern , m. , hill , r. s. , jarosik , n. , kogut , a. , komatsu , e. , limon , m. , odegard , n. , meyer , s. s. , page , l. , peiris , h. v. , spergel , d. n. , tucker , g. s. , verde , l. , weiland , j. l. , wollack , e. , & wright , e. l. 2006 , arxiv astrophysics e - prints , g. , spergel , d. n. , verde , l. , hill , r. s. , meyer , s. s. , barnes , c. , bennett , c. l. , halpern , m. , jarosik , n. , kogut , a. , komatsu , e. , limon , m. , page , l. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2003 , , 148 , 135 , n. & komatsu , e. 2006 , , 73 , 083007 , a. , spergel , d. n. , barnes , c. , bennett , c. l. , halpern , m. , hinshaw , g. , jarosik , n. , limon , m. , meyer , s. s. , page , l. , tucker , g. s. , wollack , e. , & wright , e. l. 2003 , , 148 , 161 komatsu , e. , kogut , a. , nolta , m. n. , bennett , c. l. , halpern , m. , hinshaw , g. , jarosik , n. , limon , m. , meyer , s. s. , page , l. , spergel , d. n. , tucker , g. s. , verde , l. , wollack , e. , & wright , e. l. 2003 , astrophys . j. , 148 , 119 komatsu , e. n. & spergel , d. n. 2001 , phys . rev .
d , 63 , 063002 komatsu , e. n. , spergel , d. n. , & wandelt , b. d. 2005 , astrophys .
j. , 634 , 14 komatsu , e. n. , wandelt , b. d. , spergel , d. n. , banday , a. j. , & gorski , k. m. 2002 , astrophys .
j. , 566 , 19 kovac , j. m. , leitch , e. m. , pryke , c. , carlstrom , j. e. , halverson , n. w. , & holzapfel , w. l. 2002 , nature , 420 , 772 larson , d. l. & wandelt , b. d. 2004 , astrophys .
j. , 613 , l85 linde , a. d. 1982 , phys .
lett . , b108 , 389 , d. h. & rodrguez , y. 2005 , physical review letters , 95 , 121302 , d. h. , ungarelli , c. , & wands , d. 2003 , , 67 , 023503 maldacena , j. 2003 , j. high energy phys .
, 05 , 013 , t. e. , ade , p. a. r. , bock , j. j. , bond , j. r. , borrill , j. , boscaleri , a. , cabella , p. , contaldi , c. r. , crill , b. p. , de bernardis , p. , de gasperis , g. , de oliveira - costa , a. , de troia , g. , di stefano , g. , hivon , e. , jaffe , a. h. , kisner , t. s. , jones , w. c. , lange , a. e. , masi , s. , mauskopf , p. d. , mactavish , c. j. , melchiorri , a. , natoli , p. , netterfield , c. b. , pascale , e. , piacentini , f. , pogosyan , d. , polenta , g. , prunet , s. , ricciardi , s. , romeo , g. , ruhl , j. e. , santini , p. , tegmark , m. , veneziani , m. , & vittorio , n. 2006 , , 647 , 813 mukhanov , v. f. , feldman , h. a. , & brandenberger , r. h. 1992 , phys .
, 215 , 203 mukherjee , p. & wang , y. 2004 , astrophys .
j. , 613 , 51 , l. , hinshaw , g. , komatsu , e. , nolta , m. r. , spergel , d. n. , bennett , c. l. , barnes , c. , bean , r. , dore , o. , halpern , m. , hill , r. s. , jarosik , n. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , peiris , h. v. , tucker , g. s. , verde , l. , weiland , j. l. , wollack , e. , & wright , e. l. 2006 , apjs , in press ( astro - ph/0603450 ) , h. v. , komatsu , e. , verde , l. , spergel , d. n. , bennett , c. l. , halpern , m. , hinshaw , g. , jarosik , n. , kogut , a. , limon , m. , meyer , s. s. , page , l. , tucker , g. s. , wollack , e. , & wright , e. l. 2003 , , 148 , 213 , d. s. & bond , j. r. 1990 , , 42 , 3936 . 1991 , , 43 , 1005 sato , k. 1981 , phys .
, 99b , 66 seljak , u. & zaldarriaga , m. 1996 , astrophys .
j , 469 , 437 , k. m. & zaldarriaga , m. 2006 , arxiv astrophysics e - prints , astro - ph/0612571 , d. n. , bean , r. , dore , o. , nolta , m. r. , bennett , c. l. , hinshaw , g. , jarosik , n. , komatsu , e. , page , l. , peiris , h. v. , verde , l. , barnes , c. , halpern , m. , hill , r. s. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2006 , apjs , in press ( astro - ph/0603449 ) , d. n. , verde , l. , peiris , h. v. , komatsu , e. , nolta , m. r. , bennett , c. l. , halpern , m. , hinshaw , g. , jarosik , n. , kogut , a. , limon , m. , meyer , s. s. , page , l. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2003 , , 148 , 175 starobinsky , a. a. 1982 , phys .
b. , 117 , 175 , r. 2006 , arxiv astrophysics e - prints , astro - ph/0608116 , l. , wang , l. , heavens , a. f. , & kamionkowski , m. 2000 , , 313 , 141 vielva , p. , gonzalez , e. m. , barreiro , r. b. , sanz , j. l. , & cayon , l. 2004 , astrophys .
j , 609 , 22 , l. & kamionkowski , m. 2000 , , 61 , 063504 yadav , a. p. s. & wandelt , b. d. 2005 , phys .
d , 70 , 123004
we derive an expectation value of the cubic statistics given by eq .
[ s_prim ] @xmath82 using the form of b and a as given by eq .
[ b ] and [ a ] @xmath83 which simplifies to @xmath84 where @xmath85and @xmath86 is the angular bispectrum , and @xmath87is the cmb bispectrum and can be averaged as above due to isotropy . in deriving @xmath88
we have also used : @xmath89 the theoretical primordial angular bispectrum can be written as : @xmath90 where @xmath91 using the above form of the theoretical bispectrum , @xmath92 further simplifies to @xmath93 now since @xmath94 , where @xmath95 , which is 1 when @xmath96 , 6 when @xmath97 , and 2 otherwise .
in this appendix we prove the equivalence between the optimal estimator given by @xcite and our fast estimator eqn .
( [ eq : est2 ] ) @xmath99 this is analogous to the temperature - only case @xcite . on the left hand side @xmath59 and @xmath60
run over all the 8 possible ordered combinations @xmath62 , while for the estimator on the right hand side @xmath100 and @xmath101 run only over the four unordered combinations @xmath61 .
we prove the equivalence between the optimal estimator and the fast estimator for only one combination of @xmath102 @xmath103 and @xmath104 , as the proof is same for all the combinations.the covariance matrix * cov * is obtained in terms of @xmath105 , @xmath106 , and @xmath107 ( as in equation 7 in babich and zaldarriaga ( 2004 ) ) by applying wick s theorem , @xmath108 the covariance matrix above has the form @xmath109 where @xmath100 , @xmath101 , run over @xmath61 , which after simplification gives : where @xmath111 are the @xmath112 elements of the matrix @xmath113 .
the above matrix is nothing but @xmath114 , after the additional permutations with fixed @xmath100 and @xmath101 are summed up .
for example , for @xmath115 the @xmath59 index runs over the set @xmath116 .
this completes the proof . | measurements of primordial non - gaussianity ( @xmath0 ) open a new window onto the physics of inflation .
we describe a fast cubic ( bispectrum ) estimator of @xmath0 , using a combined analysis of temperature and polarization observations .
the speed of our estimator allows us to use a sufficient number of monte carlo simulations to characterize its statistical properties in the presence of real world issues such as instrumental effects , partial sky coverage , and foreground contamination .
we find that our estimator is optimal , where optimality is defined by saturation of the cramer rao bound , if noise is homogeneous .
our estimator is also computationally efficient , scaling as @xmath1 compared to the @xmath2 scaling of the brute force bispectrum calculation for sky maps with @xmath3 pixels . for planck
this translates into a speed - up by factors of millions , reducing the required computing time from thousands of years to just hours and thus making @xmath0 estimation feasible for future surveys .
our estimator in its current form is optimal if noise is homogeneous . in future
work our fast polarized bispectrum estimator should be extended to deal with inhomogeneous noise in an analogous way to how the existing fast temperature estimator was generalized . |
it has been known since the first observations of molecular gas in star - forming clouds that the co lines exhibit supersonic line widths @xcite .
@xcite suggested that such supersonic linewidths could be produced by large - scale collapse of the molecular clouds .
in contrast , @xcite dismissed the idea of large - scale collapse with the argument that , if clouds were collapsing freely , the star formation rate in the galaxy should be a factor of 100 times larger than observed , leading to quick exhaustion of its gas content .
those authors proposed instead that the large linewidths are produced by small - scale supersonic turbulence .
since then , the scenario of supersonic molecular cloud turbulence has received much attention , and turbulent models of molecular clouds have been developed by several groups ( see * ? ? ?
* ; * ? ? ?
* and references therein ) .
however , the turbulent picture faces several problems , not the least of which is that supersonic shocks dissipate energy quickly , within one dynamical time scale , even if turbulence is magnetized @xcite .
moreover , although stellar energy has been proposed as a source of turbulence within molecular clouds , it is not clear that such feedback can maintain the identity of the clouds ; high - mass stars can easily blow apart clouds , and bipolar flows from low - mass stars are so focused that general support seems unlikely ( see review by * ? ?
thus it is far from clear - and it has not been demonstrated numerically - that stellar feedback can keep clouds near equilibrium for several dynamical timescales .
in fact , the available evidence suggest the opposite ( see * ? ? ?
the idea that molecular clouds ( mcs ) are the result of large - scale compressions of the h i diffuse medium was explored with numerical simulations of a 1 kpc squared piece of the galactic disk , representative of the solar neighborhood by @xcite and @xcite , and collected in a coherent scenario by @xcite .
these authors found that mcs can be formed rapidly and proceed to star formation almost at the same time , since the column density threshold for gravitational instability and for molecular gas formation are similar .
simultaneously , @xcite discussed the idea of large - scale compressions of the diffuse medium undergoing thermal instability , thus producing dense gas .
this led to the idea that molecular clouds in the solar neighborhood are formed in large scale compressions from the warm , diffuse , thermally unstable medium @xcite .
as these clouds accumulate mass and cool rapidly , they become jeans unstable ; at this point , gravity begins to dominate the motions , developing near equipartition between the gravitational and kinetic energies @xcite , erroneously but frequently inferred to be in `` virial equilibrium '' @xcite . as a consequence ,
the initial turbulent fluctuations in density and velocity produce hierarchical fragmentation , in which dense clumps with short free - fall times proceed to a rapid collapse , while at the same time the whole cloud contracts at a slower rate , because of its lower average density .
this model of molecular cloud formation has been subject to different observational tests . for instance
, @xcite showed that the co(1 - 0 ) observed line profiles through such modeled clouds reproduce the supersonic , turbulent line profiles of actual molecular clouds , as well as the so - called core - to - core velocity dispersion .
in other words , such global collapse of irregular structures develops internal disordered , turbulent motions at some level , and thus the large linewidths in actual mcs must be showing mainly the large - scale systematic , inhomogeneous inward motions , rather than pure turbulence .
in fact , @xcite showed that the culmination of the collapse of a medium - sized cloud produces a dense clump with large velocity dispersion , with physical properties similar to those of massive star - forming regions . in this clump ,
the velocity field is dominated by the collapse motions .
in addition , @xcite , based on observations of massive dense cores @xcite , showed that the @xcite relationship between velocity dispersion and size arises naturally if turbulent motions are the result of hierarchical and chaotic global + local collapse .
in particular , they found that the velocity dispersion should scale as @xmath0 where @xmath1 is the velocity dispersion of the gas , @xmath2 is the column density of the core , @xmath3 its size , and @xmath4 the universal constant of gravity , according to the @xcite relation recently found for molecular clouds . in the present work
we take a further step to compare models with observations .
we analyze the column density probability distribution function from models of molecular cloud formation and evolution via warm , thermally unstable h i stream collision .
the goal of the present contribution is to assess the evolution of the column density probability distribution function ( ) in these models , and compare it to the of observed molecular clouds .
this is particularly important because a number of recent observations have reported for several nearby molecular clouds , permitting more tests of theory . in section
[ sec : knowledge ] we discuss the current knowledge of volumetric and column density pdfs for interstellar gas . in
[ sec : models ] we briefly describe the models used , and in [ sec : results ] we present the time evolution of our s. in [ sec : discussion ] we present a brief discussion and in [ sec : conclusions ] we give our main conclusions .
one important tool for understanding the internal structure of molecular clouds is the volume density probability distribution ( ) . for isothermal , turbulent flows of a given mach number @xmath5 , where shocks occur stochastically and in succession , creating density fluctuations @xmath6 , the is a lognormal function , i.e. , a gaussian function in the logarithm of the density . @xcite and
@xcite explain this result as follows : in an isothermal flow , the speed of sound is spatially uniform , and a shock of intensity @xmath5 will induce a density jump @xmath7 over the mean density @xmath8 . if another shock of intensity @xmath5 arrives at the place where the density is now @xmath9 , it will induce another density jump of size @xmath10 .
thus , if the flow is stochastic , such density jumps must be spatially uniform , and a given density distribution must be obtained by a succession of multiplicative density jumps , which are additive in the logarithm .
thus , at a given position in space , each density jump is independent of the previous and the following ones , and therefore , the central limit theorem , according to which the distribution of the sum of identically - distributed , independent events approaches a gaussian , can be applied to the logarithm of the density , resulting in a lognormal density probability distribution function .
various authors suggested that the column density of turbulent , isothermal flows must also be lognormal ( e.g. , * ? ? ?
* ; * ? ? ?
however , this is only a consequence of the fact that interstellar turbulence contains the largest velocities at the largest scales , and thus , the size of the interstellar clouds is comparable to their correlation length . in this case , @xcite showed that the local values of the volume density along a single line of sight are correlated , and thus the column density might be representative of the mean value of the volume density along such line of sight .
thus , for different lines of sight , the values of the are correlated to the mean values of the , and thus , the follows the shape of the .
this would not be the case , however , if the correlation lengths of molecular clouds were small . in this case , the line of sight will contain many correlation lengths , and thus , the mean value will not be representative of the volumetric density of the line of sight . in this case
, the central limit applies over the sum of several correlation lengths , and the column density should exhibit a gaussian shape @xcite . in recent years , a number of observational studies have focused on understanding the column density pdf of nearby clouds .
for instance , @xcite used extinction data from the complete ( coordinated molecular probe line extinction thermal emission ) survey of star - forming regions , to argue that the gas in ophiuchus and perseus molecular clouds exhibits a lognormal ( see also * ? ? ? * ) .
unfortunately , the column density in this work is plotted in linear scale in the @xmath11 axis ( number of events ) , and thus , the excess in the high column density part of the can not be properly appreciated . however , that excess clearly indicates that some degree of departure of the lognormality is going on in perseus . on the other hand , @xcite presented maps and s of 14 different regions inferred from extinction measurements .
they found that the s exhibit different shapes , although the dynamical range in visual extinction @xmath12 is frequently smaller than one order of magnitude , making it difficult to draw firm conclusions about the actual shape of the for these clouds .
@xcite took this a step further , showing that the coalsack and lupus v molecular clouds , which do not have signs of star formation , exhibit lognormal s , while clouds that have already formed stars , like taurus and lupus i , do exhibit power - law like tails at large column densities .
their results were confirmed almost immediately afterwards by @xcite .
the standard interpretation of both groups is that clouds that are turbulent and non - star forming , exhibit a lognormal , while clouds that are forming stars are somehow decoupled , and exhibit a power - law tail at large column densities .
the above observational results have a clear counterpart in numerical simulations of star formation in turbulent , isothermal mcs .
several groups have shown that the s in those simulations develop power - law tails at late stages @xcite .
more recently , @xcite have shown that centrally - peaked density distributions , such as the singular isothermal sphere @xcite , or other dynamical solutions such as a pressure - free or inside - out collapse , which are all characterized by power - law density profiles , have s and s with power - law high - density tails , whose slope does in fact depend on the slope of the density profile . also , they show that numerical simulations of self - gravitating isothermal turbulence develop power - law tails as time progresses .
thus , they attribute the development of such tails in self - gravitating turbulent flows to the formation of local collapsing sites . although they do not mention it explicitly
, one can conjecture that the pdfs of the entire turbulent flow exhibit the characteristic power - law tails of the collapsing centers , because the latter are the only sites where such high densities develop .
note also that , because those authors only evolved their simulations to less than half a free - fall time , the dominant component of the kinetic energy in the simulations was still that due to the turbulent flow . on the other hand , multiphase simulations of the formation and evolution of mcs out of the convergence of warm , diffuse gas in the presence of self - gravity by @xcite and @xcite show that the dominant component of the kinetic energy in the clouds after they become dominated by self - gravity is generalized gravitational contraction . in the remainder of this paper , we show that those multiphase simulations also develop power - law tails in their s at late times , suggesting that the pdfs of mcs should not be expected to be stationary , but rather to evolve in time , as a consequence of the transition of the clouds from being turbulence dominated to being collapse dominated . as an alternative interpretation ,
@xcite have recently shown that a lognormal column density pdf does not necessarily imply that the flow is turbulent .
in particular , they have shown that simulations of individual and multiple collapses also exhibit lognormal s at early stages in their evolution .
however , it is well known that isothermal compressible turbulence develops such a lognormal pdf @xcite , thus , it is natural to ascribe the lognormal parts of pdfs of ism turbulence simulations to supersonic isothermal turbulence .
similarly , the high - density gas in the ism is characterized by supersonic linewidths and behaves very close to isothermally , suggesting that lognormal pdfs in mcs are due to this process as well . in the next section
we will show that this behavior can be understood in a unified model of molecular cloud formation and evolution , where clouds are assembled by large - scale convergent flows in the wnm , which initially produce turbulence , but gravity rapidly takes over , producing a hierarchy of nested collapsing motions , and thus , the develops a power - law wing at large column densities .
in the present section we analyze the evolution of the in numerical simulations of the formation and evolution of molecular clouds . in order to understand the evolution of the
, we briefly describe the evolution of the clouds of those simulations ( for a review , see * ? ? ?
the models consist in the collision of two large streams of warm thermally bi - stable h i gas .
the streams collide with a given inflow velocity , on top of which fluctuations in magnitude and shape are superposed .
as the streams collide , the compressed regions undergo a nonlinearly induced thermal instability , cooling down rapidly and becoming turbulent due to a combination of various instabilities @xcite .
as the cold clouds accumulate mass , they quickly become jeans unstable at various scales .
the superposition of those different centers of collapse produces hierarchy of nested collapses , which is reflected in highly supersonic line profiles @xcite that traditionally have been interpreted as evidence of turbulence . under this general scheme , several studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have performed simulations with different numerical schemes ( sph , fixed grid or amr , respectively ) , and with slightly different initial conditions . however , in spite of the differences , the formation and evolution of the modeled molecular clouds exhibit similar characteristics : initially , all modeled clouds fragment rapidly due to dynamical an thermal instabilities , as well as the initial velocity fluctuations . however , while the models without gravity remain fragmented and do not develop large density regions , the models with gravity start to collapse transversely to the inflow direction , developing local centers of collapse with large volumetric densities ( @xmath13 6 cm3 ) , as well as large column densities , ( @xmath13 cm2 ) .
as @xcite discuss , the same cloud at different evolutionary times appears as if it were different kind of objects : at earlier times , the compressed layer could be classified as a diffuse h i cloud , with low column densities and thus , very little co content ( see * ? ? ?
* ; * ? ? ?
as time goes by , the layer accumulates more mass , column densities increase , and some co begins to form . at this stage
the cloud could be identified as a `` translucent '' cloud . later in time
, larger column densities are achieved not only because of the compression of gas along the direction of the streams , but because the transverse collapse causes a rapid increase of the column density , shielding the interior of the cloud against uv photons . at this stage
, the cloud appears mostly molecular , with dense cores that are already collapsing and which a few myr later form stars .
the precise times in which each one of these stages occurs depends on several parameters , e.g. , the initial values of the density and velocity fields , the actual values of their fluctuations , the total mass of the streams , etc .
however , once the cloud has accumulated enough mass to allow for molecule formation , the first collapsed objects appear a few myr later . in the present work , in order to focus on the shape of the during the process of cloud formation , we make use of runs without stellar feedback .
we use runs hf and gf2 of the suite presented by @xcite and laf0 run by @xcite .
run hf is similar to gf2 , but without self - gravity , and is used only for reference .
run gf2 represents the evolution of a @xmath14 pc cube filled with gas at @xmath15 k and at a density of @xmath16 cm3 .
the incoming streams are ellipsoidal cylinders with a cross section of 22 pc , an inflow velocity of 7.9 km / sec , and inflow boundary conditions on the numerical box , i.e. , infinite incoming cylinders . in run laf0
, the simulation is performed using an adaptive mesh refinement numerical scheme ( art code by * ? ? ?
the box has a size of @xmath17 pc per side , filled with a gas at @xmath18 k , and a density of @xmath19 cm3 .
the streams are circular cylinders having a diameter of 32 pc , and an incoming velocity of 5.9 km / sec , and a length of 110 pc .
the inflows are thus fully contained within the box .
this means not only that they have a finite length and duration , but also that the total mass of the box in the @xcite runs is constant , while in the runs by @xcite it is not .
this result will have implications on the height of the , as we will see in [ sec : columndenistypdfs ] . in order to trigger instabilities in the compressed layer , in the hf and
gf2 runs the shock front has a sinusoidal shape , while in laf0 run , a fluctuating velocity field is added to the inflow velocity with rms amplitude of @xmath20 30% of the inflow speed , and with @xmath20 1/2 the diameter of the inflows . for more details ,
see conditions of runs gf2 and laf0 for @xcite and @xcite , respectively . figs .
[ fig : npdf - run_hf ] , [ fig : npdf - hh08 ] , and [ fig : npdf - vs ] show the s of models hf , gf2 and laf0 , respectively . for each run
, we present two extreme cases : s calculated by projecting the density field perpendicular to the inflow and one projecting along it .
all plots exhibit 16 panels , from 3 to 15 myr in figs .
[ fig : npdf - run_hf ] and [ fig : npdf - hh08 ] , and from 5.6 myr to 26.6 myr in fig .
[ fig : npdf - vs ] . in both cases , the time is measured from the beginning of the collision .
the first point to notice in all three figures is that in all the simulations , at every time step , and in both projections , there is a large number of data points at low column densities .
this is just a consequence of the setup of the simulations : the incoming flows are cylindrical , and outside these cylinders the gas is uniform , with low volume density and zero initial velocity .
the peaks in each case simply reflect the column density of the uniform background medium .
the mean density of this field is @xmath20 1 cm3 , and thus , the typical column density of this part of the box is @xmath21 cm2 for run gf2 and @xmath22 cm2 for the laf0 one .
the second point to notice is that the column densities in both projections are extreme cases because these are extreme projections . typically ,
molecular clouds must be observed at intermediate angles . in any case , the initial s for the projection along the flow ( plane @xmath23 ) are narrower than the projections perpendicular to the flow ( planes @xmath24 or @xmath25 , not showed here ) .
this is because the clouds are formed in the @xmath23 plane , and hence the cloud is thinnest along the @xmath26-direction . in all three figures
we further show lognormal and/or power - law fits as follows : fig .
[ fig : npdf - run_hf ] shows only lognormal fits ( dotted line ) at every timestep , since the corresponding run is purely hydrodynamical .
note that the pdf increases in height as time goes on .
this is a consequence of the inflow boundary conditions of the models from @xcite , which imply that the total mass in the box increases in time . in figs .
[ fig : npdf - hh08 ] and [ fig : npdf - vs ] , on the other hand , we show the lognormal fits from @xmath27 5.3 myr to @xmath27 9.1 myr and from @xmath27 9.8 myr to @xmath27 16.8 myr , respectively .
the final times of these intervals correspond to the time at which the cloud start collapsing more vigorously .
however , for reference , we keep the first fit until the end of the simulation ( dot - dashed line ) .
we note that in both panels of fig . [
fig : npdf - run_hf ] the s exhibit a lognormal - like wing at large column densities ( dotted line ) since early on in the simulation s evolution , and maintain this shape for the rest of the evolution .
the lognormal wing at large column densities is due to the density fluctuations produced by the turbulence induced by a combination of nonlinear thin shell , thermal and kelvin - helmholtz instabilities in the compressed layer @xcite .
however , it is worth noting that this lognormal part is quite narrow , with column densities spanning less than two orders of magnitude , from @xmath28 cm2 to 10 cm2 , for the entire duration of the evolution .
this is due to the fact that the turbulence generated by the instabilities is not very strong , in agreement with previous results @xcite .
the case with self - gravity is quite different ( see figs .
[ fig : npdf - hh08 ] and [ fig : npdf - vs ] ) .
we divide the behavior in two cases : the lognormal and the power - law behaviors .
we first note that , after some transients , the dense column density region develops a narrow lognormal shape , indicated by the dotted line ( which we mantain through all frames for clarity ) .
however , as time goes on , gravity modifies the lognormal shape , first by making it wider . in order to show this widening more clearly , in fig.[fig : sigma ] we plot the evolution of the standard deviation @xmath29 of the lognormal fits as a function of time for the four runs with self - gravity . from this figure
it is clear that the lognormals tend to increase in width as the collapse proceeds .
this behavior has been recently reported by @xcite .
finally , at late times , gravity runs away and column density grows several orders of magnitude in a few myr . as a consequence , the develops a power - law tail at large column densities .
time evolution of the standard deviation of the lognormal function fitted in the two projections of the two runs with self - gravity . as can be seen , in all cases @xmath29 increases with time , implying that gravity makes wider the lognormal of a purely turbulent gas . ]
the s shown in figs .
[ fig : npdf - hh08 ] and [ fig : npdf - vs ] are reminiscent of the variety of s for different clouds shown by @xcite .
in fact , as shown by @xcite , every cloud may have its own slope , depending not only on the physical conditions of the cloud , but also on the resolution at which the cloud itself has been observed .
furthermore , by comparing the direction of the projections , we note that the slope depends on the orientation of the cloud : although in models gf2 the slope in both projections is almost the same ( see fig [ fig : npdf - hh08 ] ) , models laf0 ( fig [ fig : npdf - vs ] ) show that the slope is steeper when we observe the cloud along its longer dimension .
this effect occurs because if the cloud is mostly aligned with the line of sight , the observed column densities will be larger than the column densities obtained if the cloud is mostly on the plane of the sky .
this result contrasts with the suggestion by @xcite that all the s should tend to evolve towards a single value .
these authors used a model of a spherical cloud , and thus every projection is the same .
however , real clouds are far from round , and thus one can not expect an asymptotic single value of the slope of the power - law tail of the . also , it is noteworthy that the other run with slightly different perturbations in @xcite , ( run gf1 , not shown ) , which has a smaller perturbation in the shock front , has an evolution which is basically indistinguishable from the one presented here for model gf2 .
similarly , the run with smaller amplitude fluctuations ( saf0 , also not shown ) in @xcite , follows an evolution indistinguishable from model laf0 . however , the precise time in which each model @xcite start collapsing are different : @xmath30 myr for the @xcite runs , and @xmath31 myr for the @xcite runs . the difference seems to be due to the free - fall time in each run : the compressed region in the @xcite runs has 47% of the volume of the compressed region in @xcite , but its mass flux is 5 times larger .
thus , the density in the compressed region is 2.23 times larger in the @xcite run , which in turn implies a free - fall timescale 2/3 times smaller .
this result suggests that , to first order , the main parameters that define the time of collapse are the velocity and the density of the incoming field .
indeed , a parameter study by @xcite has found widely different star formation efficiencies and times of the onset of star formation as parameters such as the inflow velocity and the amplitude of the background turbulent fluctuations are varied . in both runs , after the development of the lognormal , a clear power - law tail develops at high column densities .
this shape transition marks also the beginning of the accelerated large - scale collapse ongoing simultaneously with the small - scale local collapses , and occurs in few myr ( @xmath32 myr ) , regardless of the particular setup of the runs .
it seems thus that once the cloud has achieved enough shielding to form molecules , the cloud proceeds to collapse in few myr , as pointed out by @xcite .
the rapidity of this transition explains why most of the molecular clouds in the solar neighborhood exhibit young star forming regions ( few myrs ) , and there are only few clouds with no signs of star formation ( coalsack , lupus v , see * ? ? ?
thus , we conclude that as global@xmath33local collapses proceed , the number of high column density points in the s increases , causing a high tail .
as pointed out by @xcite and @xcite , in the solar neighborhood the pressure due to self - gravity becomes importat at about the column density necessary to shield the compressed region against uv radiation ( cm2 ) , allowing the formation of molecules .
this situation corresponds to the 3rd row in the rightmost column of fig . 1 in @xcite , and corresponds to the moment in which the cloud has already started to collapse in the direction perpendicular to the shock . at this moment , the densest cores of the cloud reach column densities of cm2 .
@xcite , @xcite and @xcite , argue that the different shapes of the observed s must be due to a change in the governing physical processes in molecular clouds .
in particular , @xcite argue that most of their s can be fitted by a lognormal distribution , plus a power - law wing , the first one for low column densities , which corresponds to the gas that is not forming stars , and the second one for the gas which is forming stars .
our results show that these two regimes ( quiescent vs. active star formation ) may very well be part of a single evolutionary process , as depicted by @xcite . in the early stages of the molecular cloud ,
the h i colliding streams form the molecular gas and provide part of the supersonic turbulent kinetic energy observed in the line profiles . during this stage , the is lognormal , and
most of this stage occurs in the atomic phase @xcite , during which the cloud does not form stars .
later on , hierarchical and chaotic gravitational contraction takes over , tending to produce a power - law tail in the s. the initial manifestation of this effect is that the lognormal appears to become wider .
it is important to mention that gravity is required in order to achieve the 1.52 orders of magnitude in the spread of the column density values , since pure turbulence produces only very narrow lognormal s. the contribution of self - gravity in increasing the column density helps the cloud to achieve values high enough to rapidly form molecular gas , as pointed out by @xcite .
a few myr after the molecules have formed , the densest cores collapse and the first stars begin to form .
although we agree with the general scheme depicted by @xcite , we stress that the contribution of self - gravity is indispensable to achieve high values of the column density .
that is , self - gravity is not only important in driving the collapse of gravitationally unstable clumps , but also in _ forming _ such clumps , which would not be produced by turbulence alone @xcite . during the later stages ,
the primary source of `` turbulent '' motions ( i.e. , supersonic linewidths ) in the molecular cloud is gravity as well , with local mass concentrations that render the velocity field more locally complex @xcite .
using numerical models of the formation and evolution of molecular cloud from the diffuse interstellar medium , we have confirmed the suggestion by @xcite that the transits from a lognormal shape at early times , when the kinetic energy in the clouds is dominated by their initial turbulence ( produced by various instabilities in the compressed layer that becomes the cloud ) , to having a power - law tail at high @xmath34 at late times , when their kinetic energy is dominated by gravitational contraction , which culminates with the formation of stars .
our results thus explain naturally recent observations showing that non - star - forming clouds exhibit lognormal s , while star - forming clouds exhibit power - law tails at high densities , understanding the result as the consequence of the transition of the clouds from a more diffuse , turbulence - dominated regime to a denser , star - forming , collapsing one .
we also have shown that the slope of the power - law tail produced by gravity in the does not have a unique value , but depends on the orientation of the cloud with respect to the line of sight .
this work has received partial support from grants unam / dgapa in110409 to jbp , conacyt 102488 to evs , nsf ast-0807305 to lh and fh , and has made extensive use of the nasa s astrophysics data system abstract service . | it has been recently shown that molecular clouds do not exhibit a unique shape for the column density probability distribution function ( ) . instead
, clouds without star formation seem to possess a lognormal distribution , while clouds with active star formation develope a power - law tail at high column densities . the lognormal behavior of the has been interpreted in terms of turbulent motions dominating the dynamics of the clouds , while the power - law behavior occurs when the cloud is dominated by gravity . in the present contribution we use thermally bi - stable numerical simulations of cloud formation and evolution to show that , indeed , these two regimes can be understood in terms of the formation and evolution of molecular clouds :
a very narrow lognormal regime appears when the cloud is being assembled .
however , as the global gravitational contraction occurs , the initial density fluctuations are enhanced , resulting , first , in a wider lognormal , and later , in a power - law .
we thus suggest that the observed of molecular clouds are a manifestation of their global gravitationally contracting state .
we also show that , contrary to recent suggestions , the exact value of the power - law slope is not unique , as it depends on the projection in which the cloud is being observed .
[ firstpage ] ism : general clouds kinematics and dynamics turbulence stars : formation |
we introduce an updated and largely extended version of therminator @xcite , the thermal heavy ion generator , created to carry out the statistical hadronization in relativistic heavy - ion collisions .
numerous successful analyses have been performed with the help of our code over the last few years @xcite , contributing to better understanding of the involved evolution of the hot and dense system created in ultra - relativistic heavy - ion collision .
therminator 2 is a monte carlo generator written in c++ and using the standard cern root @xcite environment . that way ,
apart from model applications , the code can be easily adapted for purposes directly linked to experimental data analysis , detector modeling , or estimates for the heavy - ion experiments at rhic , lhc , sps , fair , or nica .
originally , therminator was designed to perform the hadronic freeze - out on simple boost - invariant hypersurfaces , such as the cracow single - freeze - out model @xcite and the blast - wave model @xcite . meanwhile ,
the code has evolved into a versatile tool , where the freeze - out profile and the expansion velocity field of any shape can be implemented , allowing application to all approaches based on statistical hadronization on a specified hypersurface .
the present functionality of the code includes the following new features : 1 .
an implementation of any shape of the freeze - out hypersurface and the expansion velocity field is possible now , including the 2 + 1 and 3 + 1 dimensional profiles , in particular those generated externally with various codes for perfect hydrodynamics @xcite .
2 . the hypersufraces may have space - time dependent thermal parameters , which allows for studies of non - boost - invariant systems .
in particular , one may depart significantly from the mid - rapidity region to the fragmentation regions where the baryon chemical potential becomes large @xcite .
the package includes a library of `` standard '' sets of hypersurfaces and velocity profiles , which describe the au+au data @xcite at the highest rhic energy @xmath0 gev for various centralities .
we also provide the hypersurfaces and velocity profiles anticipated @xcite for the lhc pb+pb collisions at @xmath1 tev , which prior to studies with the real data can be used for the modeling of detectors .
4 . a separate code , the femto - therminator ,
is provided to carry out the analysis of the femtoscopic correlations .
these correlations are an important source of complementary information concerning the size and expansion of the system , which should be reproduced in a realistic description of the data @xcite .
we also provide several scripts that carry out auxiliary tasks : 1 . counting the number of particles flowing back into the fireball , thus violating causality ( for realistic freeze - out profiles this number is negligible , below 1% ) .
an estimate of the number of elastic collisions ( trajectory crossings ) after the freeze - out . for realistic cases
this number is small ( @xmath4 for pions scattering off other pions ) , supporting the single - freeze - out approximation .
3 . visualization of the results : the particle @xmath2-spectra , the elliptic flow coefficient @xmath5 , or the hbt correlation radii .
we have also made an effort to make the package more user - friendly , providing numerous examples of usage in typical situations .
other codes implementing the thermal description of hadronization have also been developed recently , to mention share @xcite and thermus @xcite , computing the particle abundances , dragon @xcite , a monte carlo generator of particle production from a fragmented fireball , or hydjet++ @xcite , a heavy - ion event generator with hydrodynamics and jets .
a code similar to therminator , fast hadron freeze - out generator , has been presented in refs .
for the detailed description of the physics behind therminator 2 the reader is referred to the original paper @xcite . here
we only wish to stress that the statistical approach to heavy - ion collisions has proved to be very useful for a very broad range of observables . with few physical input parameters , such as the temperature , chemical potentials , size , and the velocity of the collective flow ,
the models describe the observed particle abundances @xcite , the transverse - momentum spectra @xcite , the balance functions @xcite , the elliptic flow coefficient @xcite , and femtoscopic observables @xcite . a uniform description of the spectra , the elliptic flow coefficient @xmath5 , and the hbt data achieved in @xcite explained for the first time the so - called `` rhic hbt puzzle '' .
the key element of the success behind the statistical approach is the inclusion of the full list @xcite of hadronic resonances , whose number grows rapidly according to the hagedorn hypothesis @xcite . at the rather high temperature of the freeze - out ,
@xmath6 mev , the resonances contribute very significantly to the observables .
therminator 2 uses the particle data tables in the universal form originally included in the share @xcite package , and incorporates the four- * * * * and three- * * * resonances .
all particles are created at the freeze - out hypersurface according to the cooper - frye formula @xcite .
the two- and three - body decays of resonances , proceeding in cascades , produce the stable particles observed in the detectors .
therminator offers the full space - time information on positions and velocities of the produced particles . as a monte carlo event generator written in c++ in the cern root @xcite environment ,
therminator can be straightforwardly interfaced to the standard software routinely used in the data analysis of relativistic heavy - ion collisions at the lhc , rhic , sps , fair or nica .
the inclusion of experimental acceptance and kinematic cuts poses no difficulty in a monte carlo generator .
our starting point for the analysis of different models is the famous cooper - frye formula @xcite .
it expresses the number of hadrons being produced on the freeze - out hypersurface @xmath7 by the following integral @xmath8 where @xmath9 \pm 1 \right\}^{-1 } \label{eqn_pdotu0}\ ] ] is the phase - space distribution function of the particles ( the stable ones and resonances ) , and @xmath10 is the three - dimensional element of the freeze - out hypersurface .
the latter may be calculated with the help of the formula known from the differential geometry @xcite @xmath11 where @xmath12 is the levi - civita tensor ( @xmath13 ) and the variables @xmath14 , @xmath15 , and @xmath16 are used to parametrize the three - dimensional freeze - out hypersurface submerged in the four - dimensional minkowski space .
the quantity @xmath10 may be interpreted as the four - vector normal to the hypersurface with the norm equal to the `` volume of the hypersurface element '' . explicitly , @xmath17 while the remaining components are obtained via cyclic permutations of @xmath18 , @xmath19 , @xmath20 , and @xmath19 . in the following subsections
we shall explicitly provide @xmath21 in the form appropriate for various considered cases .
the formulas will be directly used to generate the hadrons with the monte - carlo method . .
, scaledwidth=40.0% ] the quantities appearing in the expressions below have the following physical interpretation : * thermodynamic quantities on the freeze - out hypersurface : * * @xmath22 - temperature , * * @xmath23 - baryon chemical potential , * * @xmath24 - isospin chemical potential , * * @xmath25 - strange chemical potential , * * @xmath26 - charmed chemical potential , * space - time coordinates : * * @xmath27 - longitudinal proper time , * * @xmath28 - distance in the transverse plane , * * @xmath29 - azimuthal angle , * * @xmath30 - space - time rapidity , * coordinates and parameters used to specify the freeze - out hypersurface : * * @xmath31 - the initial proper time for hydrodynamics , * * @xmath32 - the final proper time , * * @xmath33 - angle in @xmath34 plane , see fig . [ fig:2d ]
* * @xmath35 - angle between the @xmath36-axis and the direction determined by the space time point ( @xmath31,0,0,0 ) and the point on the freeze - out hypersurface , see fig .
[ fig:3d ] , * * @xmath37 - distance from the space - time point ( @xmath31,0,0,0 ) to the point on the freeze - out hypersurface , see fig .
[ fig:3d ] , * flow characteristics of the fluid element : * * @xmath38 , @xmath39 - transverse components of the four - velocity , * * @xmath40 - magnitude of the transverse velocity , * * @xmath41 - azimuthal angle of the transverse velocity , * * @xmath42 - rapidity of the fluid element , * properties of a particle generated on the freeze - out hypersurface : * * @xmath43 - spin , * * @xmath44 - mass , * * @xmath2 - transverse momentum , * * @xmath45 - transverse mass , * * @xmath46 - azimuthal angle of the transverse momentum , * * @xmath47 - rapidity , in the case of freeze - out hypersurfaces obtained in the framework of the perfect - fluid hydrodynamics the distribution function in ( [ cp1 ] ) is the fermi - dirac or bose - einstein distribution and the number of hadrons produced is given by the formula @xmath48 ) and ( [ 3d - par2 ] ) . , width=283 ] first , we consider the most general case where both the freeze - out hypersurface and the hydrodynamic flow at freeze - out are obtained from the 3 + 1 perfect - fluid hydrodynamic code , with no symmetries assumed .
in this situation the freeze - out hypersurface is parametrized in the following way @xcite , @xmath49 where @xmath50 is a function of the parameters @xmath51 , and @xmath35 , which is obtained numerically from the hydrodynamic code , see fig .
[ fig:3d ] .
the parameter @xmath52 is a scale used to change the dimensionless space - time rapidity @xmath36 into a dimensional quantity ( for example one may choose @xmath53 ) , namely @xmath54 in addition , we express the particle four - momentum and the fluid four - velocity in terms of particle and fluid rapidities , respectively , which leads to the expression @xmath55 .
\label{3d - pu}\ ] ] the calculation of the integration measure in the cooper - frye gives @xmath56 \nonumber\\ & & + \left(d \cos\theta + \frac{\partial d}{\partial\theta } \sin\theta \right ) \sin\theta \sinh\left(y - \frac{d \cos\theta}{\lambda}\right ) \biggl .
\biggr\ } .
\label{3d - sp1}\end{aligned}\ ] ] using ( [ 3d - par2 ] ) in ( [ 3d - sp1 ] ) one obtains a more compact form @xmath57 .
\label{3d - sp2}\end{aligned}\ ] ] for boost - invariant systems , considered in approximate treatment of the heavy - ion collisions , the fluid rapidity is equal to space - time rapidity , @xmath58 , and the function @xmath50 depends only on @xmath29 and @xmath33 . in this case
we find @xcite @xmath59 and @xmath60 .
\label{2d - sp}\end{aligned}\ ] ] in this section we discuss one of the most popular parameterizations of the freeze - out hypersurface the blast - wave model . in its standard form ,
the model is boost - invariant and cylindrically symmetric .
moreover , the assumption is made that the freeze - out happens at a constant value of the proper time @xmath61 in order to get a broader applicability , we generalize this condition to the formula @xmath62 where @xmath63 is a constant that describes the slope of the freeze - out curve in the minkowski space , see fig . [
fig : model - a ] . with @xmath64 ( @xmath65 )
we may consider the freeze - out scenarios where the outer parts of the system freeze - out later ( earlier ) .
of course , with @xmath66 we reproduce the standard blast - wave parametrization . in this case
we use @xmath67 and @xmath68 the freeze - out conditions defined by eq .
( [ model - a2 ] ) with different values of @xmath63 were studied in @xcite .
another boost - invariant and cylindrically symmetric freeze - out model @xcite assumes that the hypersurface is defined by the condition , @xmath69 see fig .
[ fig : hubble ] , and the fluid four - velocity is proportional to the space - time position @xmath70 note that the physical interpretation of the parameter @xmath32 is now slightly different from that used in the previous section .
the calculation of the volume element of the freeze - out hypersurface shows that it is proportional to @xmath71 , as in the model of siemens and rasmussen @xcite , so we may write @xmath72 and @xmath73 the two- and three - body resonance decays are implemented in exactly the same manner as in the original version of the code @xcite .
the format of the tables from share @xcite is presented in appendix [ appa ] .
therminator is written in c++ and uses the cern root environment _ an object - oriented data analysis framework _
@xcite .
, width=396 ] the code consists of several modules performing subsequent tasks .
first , the information on particles and their decays is read by parcer.cxx and stored by particledb.cxx .
then eventgenerator.cxx conducts the generation of events , which consists of several tasks .
integrator.cxx computes the average multiplicity of primordial particles corresponding to a given model and parameters .
next , event.cxx generates the event , first producing the primordial particles with the multiplicity given by the poisson distribution from integrator.cxx , and then calling particledacayer.cxx to carry out the decays of resonances .
finally , the event is written to the output file .
the corresponding block structure is shown in fig .
[ files : block_events ] . the organization of the input and output files is shown in fig .
[ files : block ] . after unpacking
the therminator 2 package in the directory therminator2 one obtains the directory structure presented in fig.[files : dir_struct ] .
, width=132 ] the main directory of therminator 2 contains the following files : * doxygen doxygen @xcite configuration , * events.ini therminator 2 configuration used by therm2_events , * femto.ini therminator 2 configuration used by therm2_femto , * hbtfit.ini therminator 2 configuration used by therm2_hbtfit , * makefile make configuration , * manual.html reference manual in html format ( created with make doc ) , redirection to doc / html / index.html , * manual.pdf reference manual in pdf format ( created with make doc ) , symbolic link to doc / latex / refman.pdf , * runtorque.sh bash shell script for torque @xcite , * runall.sh bash shell script for the full calculation , * runevents.sh bash shell script for generating events , * runfigure.sh bash shell script for generating figures , * runhbt.sh bash shell script for the femtoscopic analysis , * therm2_events program generating events ( created with make ) , * therm2_femto program for two - particle correlations ( created with make ) , * therm2_hbtfit program for hbt analysis ( created with make ) .
the other directories are : therminator2/build / doxygen/ contains doxygen input files used to generate the reference manual , therminator2/build / include/ contains header files ( .h ) with the declarations of classes and structures , as well as the general declaration file thglobal.h , therminator2/build / obj/ contains the compiled source files ( .o ) and the binaries therm2_events and therm2_femto ( created with make ) , therminator2/build / src/ contains the source code files for the classes ( * .cxx ) and the programs therm2_events.cxx and therm2_femto.cxx , therminator2/doc/ contains the doxygen reference manuals in the html and latex formats ( created with make doc ) , therminator2/events/ contains the generated event files , figures , correlation function files , etc .
( the contents of the events/ directory and the subdirectory structure depends on the selected model and its settings ) .
the default subdirectories for each model are as follows : * therminator2/event / krakow/ the krakow single freeze - out model , * therminator2/event / blastwave/ the blast - wave model , * therminator2/event / bwa/ the blast - wave a - class model , * therminator2/event / lhyquid2dbi-*/ the lhyquid 2 + 1d boost - invariant hydro , * therminator2/event / lhyquid3d/ the lhyquid 3 + 1d hydro .
the files stored in the events/*/ subdirectory : * event*.root event file , * fmultiplicity_.txt primordial particle multiplicity , * fig_.eps figure in eps format , * fig_.png figure in png format , * fig_.xml figure data in xml format , * fig_.c root macro to generate a figure , * femto*.root two particle correlation function , * hbtfit*.root fit to the two particle correlation function , * events.ini event generator configuration file , * femto.ini two - particle correlation configuration file , * hbtfit.ini hbt fit configuration file , * model_config.ini freeze - out model configuration file ( the file name depends on the chosen model , e.g. lhyquid2dbi.ini ) .
file naming conventions are as follows : the event*.root files are always stored in a sequence beginning with event000.root , and subsequently with the file counter incremented by one .
the fmultiplicity_.txt file has an eight - character hex - base number representing the crc-32 identification number calculated from the model parameters , e.g. , fmultiplicity_10d57d490.txt for the default case of rhic with the centrality 20 - 30% .
the fig_. * files have the name inherited from the corresponding root macros ( with the word `` figure '' shortened to `` fig '' ) .
the femto*.root and fmfit*.root filenames depend on the particle pair type , average pair momentum , and the inclusion of particles form resonances , e.g. , femtopipi0200a.root for the pion - pion pairs with average @xmath3 = 200 mev , and with resonance decays idicated with `` a '' .
therminator2/fomodel/ contains the freeze - out model configuration files ( .ini ) and freeze - out hypersurfaces data in the xml format ( * .xml ) .
the five freeze - out model configuration files are supplied : * blastwave.ini for blast - wave model , * bwa.ini for modified blast - wave a - class model , * krakow.ini for krakow single freeze - out model , * lhyquid2dbi.ini for 2 + 1 boost - invariant hydrodynamics , * lhyquid3d.ini for 3 + 1 hydrodynamics .
the freeze - out hypersurface input files for rhic at @xmath74 gev and various centralities have the following names : * lhyquid2dbi / rhicauau200c0005ti500ti025tf145.xml ( for @xmath75 = 05 % ) , * lhyquid2dbi / rhicauau200c0510ti491ti025tf145.xml ( 510 % ) , * lhyquid2dbi / rhicauau200c1020ti476ti025tf145.xml ( 1020 % ) , * lhyquid2dbi / rhicauau200c2030ti455ti025tf145.xml ( 2030 % ) , * lhyquid2dbi / rhicauau200c3040ti429ti025tf145.xml ( 3040 % ) , * lhyquid2dbi / rhicauau200c4050ti398ti025tf145.xml ( 4050 % ) , * lhyquid2dbi / rhicauau200c5060ti354ti025tf145.xml ( 5060 % ) , * lhyquid2dbi / rhicauau200c6070ti279ti025tf145.xml ( 6070 % ) .
the name encodes the colliding nuclei , the collision energy ( in mev ) , the centrality class , the initial central temperature ti ( in mev ) , the starting time used in hydrodynamics ti ( in fm / c ) , and the freeze - out temperature tf ( in mev ) . in an analogous way ,
the freeze - out hypersurface input files for the lhc collisions at @xmath76 tev and various centralities are : * lhyquid2dbi / lhcpbpb5500c0005ti500ti100tf145.xml ( for @xmath75 = 05 % ) , * lhyquid2dbi / lhcpbpb5500c0510ti500ti100tf145.xml ( 510 % ) , * lhyquid2dbi / lhcpbpb5500c1020ti500ti100tf145.xml ( 1020 % ) , * lhyquid2dbi / lhcpbpb5500c2030ti500ti100tf145.xml ( 2030 % ) , * lhyquid2dbi / lhcpbpb5500c3040ti500ti100tf145.xml ( 3040 % ) , * lhyquid2dbi / lhcpbpb5500c4050ti500ti100tf145.xml ( 4050 % ) , * lhyquid2dbi / lhcpbpb5500c5060ti500ti100tf145.xml ( 5060 % ) , * lhyquid2dbi / lhcpbpb5500c6070ti500ti100tf145.xml ( 6070 % ) .
therminator2/macro/ contains the root macro files for generating figures in the event analysis : * figure_disteta.c distribution in pseudorapidity , * figure_distmt.c distribution in transverse mass , * figure_distpt.c distribution in transverse momentum , * figure_distpt_exotic.c distribution in transverse momentum of less abundant particles ( e.g. , those with strangeness ) , * figure_distpt_pion.c pions anatomy in transverse momentum ( contributions from individual resonance decays to pion spectra ) , * figure_distpt_proton.c protons anatomy in transverse momentum ( contributions from individual resonance decays to proton spectra ) , * figure_distrap.c distribution in rapidity , * figure_distrap_prim.c primordial - particle distribution in rapidity , * figure_hsret.c primordial - particle distribution in @xmath77 , * figure_hsxyt.c primordial - particle distribution in @xmath78 , * figure_xpx.c position - momentum correlation , @xmath79 , * figure_v2pt.c differential elliptic flow at midrapidity , * figure_v4pt.c differential @xmath80 coefficient at midrapidity , * figure_v2rap.c integrated elliptic flow vs. rapidity .
other root macros used by the figure*.c files are : * events2chain.c reads event files , * hist2xml.c converts histogram data into the xml format , * model2legend.c creates figure legends displaying the model parameters .
therminator2/share contains the share @xcite database files : * decays.data particle decay channels , * particles.data particle properties .
therminator2/addons contains an auxiliary program ( estimate of the elastic rescattering ) : * coll.mk makefile , * collcount.cxx program estimating elastic collisions per pion . .... # ----------------------|-------------------------------| # freezeoutmodel | model parameter file | # ----------------------|-------------------------------| # krakowsfo | ./fomodel
/ cracow.ini | # blastwave | ./fomodel / blastwave.ini | # bwavt | ./fomodel / bwa.ini | # bwavtdelay | ./fomodel
/ bwa.ini | # bwavlinear | ./fomodel / bwa.ini | # bwavlineardelay | ./fomodel / bwa.ini | # bwavlinearformation | ./fomodel
/ bwa.ini | # lhyquid3d | ./fomodel
/ lhyquid3d.ini | # lhyquid2dbi | ./fomodel
/ lhyquid2dbi.ini | # ----------------------|-------------------------------| # name of the freeze - out model # available : see table above # default : lhyquid2dbi freezeoutmodel = lhyquid2dbi # custom freeze - out model ini file [ not used by default ] # default : # freezeoutmodelini = ./fomodel / lhyquid2dbi.ini [ event ] # number of events to generate # default : 5000 numberofevents = 5000 # event output file format # available : root , root&text , text # default : root eventfiletype = root [ primordial ] # distribution of primordial particles multiplicity # available : poisson # default : poisson multiplicitydistribution = poisson # number of samples used in determination of primordial multiplicity and max .
integrand value # default : 5000000 integratesamples = 5000000 [ random ] # start each event with a new random seed taken from current time ( 1 ) or do a constant seed ( 0 ) # default : 1 randomize = 1 [ directories ] # directory with share input files # default : ./share/ sharedir = ./share/ # directory with freeze - out model parameter files # default : ./fomodels/ freezeoutdir = ./fomodel/ # directory with root macro files * .c # default : ./macro/ macrodir = ./macro/ # directory to write the events # default : ./events/ eventdir = ./events/ [ logging ] # log file - save information on number of events runed , destination and time # default : therminator.log logfile = therminator.log .... the main configuration file events.ini ( table [ evtab ] ) is used by therm2_events program to select the model , number of generated events , type of output , file and directory names , etc .
depending on the value of freezeoutmodel , the program reads additional information from an appropriate model settings file .ini .
the file is located in the _ freezeoutdir _ directory .
the model configuration file ( table [ tabmod ] ) contains the parameters of a particular model , e.g. , the configuration file for the default * freezeoutmodel = lhyquid2dbi * is fomodel / lhyquid2dbi.ini .
the variable integratesamples , used to obtain the integrand ranges and the particle average multiplicities , is set to a large value ( 5000000 by default ) .
this leads to accurate results , but increases the running time of the first stage of therm2_events .
the user may decrease this value , however , care is needed in the inspection of results . a very large value of numberofevents , above 20000 ,
is needed to achieve stable hbt results . a low value ( about 1000 )
is sufficient to obtain the @xmath2 spectra .
of course , the running time is proportional to numberofevents . ....
# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # lhyquid 2 + 1 boost invariant # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # rapidity range # default : 4.0 rapprange = 4.0 # spatial rapidity range # default : 8.0 rapsrange = 8.0 # location of the hydro - code output containing the definition of the hypersurface , # velocity profile , thermodynamic properties etc .
# default : lhyquid2dbi / rhicauau200c2030ti455ti025tf145.xml freezefile = lhyquid2dbi / rhicauau200c0005ti500ti025tf145.xml # subdirectory to store events of this model .
# if option is hashed the eventsubdir will be made form freezefile name ( replace ' / ' with ' - ' # and remove " .xml " ) # default : lhyquid2dbi - rhicauau200c2030ti455ti025tf145/ # eventsubdir = lhyquid2dbi - rhicauau200c2030ti455ti025tf145/ ....
the particles are stored in the computer memory as collections corresponding to one event .
once the whole event is generated ( all primordial particles are produced and all unstable particles have been sequentially decayed ) , the event is written down to the output file and erased from the memory .
there are two types of output : plain ascii text and root file .
the plain text output has the advantage of having the complete information and being easily readable by a human - being or a computer program .
it is , however , not efficient in terms of disk - space usage and data - access time .
the structure of the plain text file is the following : * the first two lines form the file header containing a description of the event_entry , see further in text . *
each event in the output file consists of an event header containing the event_id ( a unique number generated with the crc-32 algorithm from the system time and event sequence number ) and number of particles in the event . * the lines that follow the event header are the particle entries .
each particle is represented by a single line .
.... # therminator 2 text output #
< event_entry > eid fathereid pid fatherpid rootpid decayed mass e px py pz t x y z</event_entry > # < event_id>0x65679f6</event_id > # < no_of_particles>9684</no_of_particles > 0 -1 9001 9001 9001 0 2.350000e+00 5.423351e+00 5.105272e-01 8.400604e-01 -4.787891e+00 3.012708e+01 3.797786e+00 1.937673e+00 -2.789553e+01 1 -1 5218 5218 5218 1 2.250000e+00 5.035699e+00 1.177745e+00 -3.007674e-01 -4.337997e+00
2.272015e+01 2.773765e+00 -1.740429e+00 -1.985397e+01 .... * * eid * , the particle sequence number in the event , * * fathereid * , the sequence number of the parent , * * pid * , particle data group ( pdg ) identification number , * * fatherpid * , pdg identification number of the parent , * * rootpid * , pdg identification number of the original primordial particle starting the decay sequence , * the decay flag ( 1 if particle has decayed , 0 if not ) , * mass and components of the four momentum : * e * , * px * , * py * , and * pz * ( in gev ) , * space - time coordinates of the creation point : * t * , * x * , * y * and * z * ( in fm / c ) . the information on * eid * , * fathereid * , together with * pid * and * fathereid * can be used to trace back the resonance cascade from the last stable particle up to its primordial particle born on the freeze - out hypersurface
. particles with * * are primordial particles with * pid = fathereid = rootpid*. the tree called * particles * consists of one branch named * particle*. the branch is filled with particle entries ( leaves ) one leaf per particle . this tree stores information on particles from 500 events ( by default ) .
the particle leaf consist of a class called * particlecoor * that stores the following particle properties : a. kisiel et al . , comput .
174 ( 2006 ) 669 , nucl - th/0504047 .
w. broniowski et al .
b635 ( 2006 ) 290 , nucl - th/0510033 .
s. pratt and d. schindel , aip conf .
828 ( 2006 ) 430 , nucl - th/0511010 .
a. andronic , p. braun - munzinger and j. stachel , nucl .
a772 ( 2006 ) 167 , nucl - th/0511071 .
a. kisiel , w. florkowski and w. broniowski , phys . rev .
c73 ( 2006 ) 064902 , nucl - th/0602039 .
m. chojnacki and w. florkowski , phys .
c74 ( 2006 ) 034905 , nucl - th/0603065
. w. florkowski et al .
, acta phys .
b37 ( 2006 ) 3381 , nucl - th/0609054 . b. biedron and w. broniowski , phys .
c75 ( 2007 ) 054905 , nucl - th/0610083 .
o. utyuzh , g. wilk and z. wlodarczyk , braz .
37 ( 2007 ) 708 , hep - ph/0610408 .
v.v . begun et al . ,
c76 ( 2007 ) 024902 , nucl - th/0611075 .
o. utyuzh , g. wilk and z. wlodarczyk , phys .
d75 ( 2007 ) 074030 , hep - ph/0702073 .
d.a . brown et al .
c76 ( 2007 ) 044906 , 0705.1337 .
b. baeuchle , ( 2007 ) , 0706.4252 .
phenix , r. vertesi , ( 2007 ) , 0706.4409 .
p. danielewicz , ( 2007 ) , 0707.0377 .
. begun , phys .
atom . nucl .
71 ( 2008 ) 1813 , 0711.2912 .
b. tomasik et al .
, acta phys .
supp . 1 ( 2008 ) 513 , 0712.0563 .
m. chojnacki et al .
c78 ( 2008 ) 014905 , 0712.0947 .
p. bozek , phys .
c77 ( 2008 ) 034911 , 0712.3498 .
phenix , s. afanasiev et al .
( 2008 ) 232301 , 0712.4372 .
y. akiba , prog .
suppl . 174
( 2008 ) 88 .
w. broniowski et al .
( 2008 ) 022301 , 0801.4361 .
w. florkowski et al .
, acta phys .
b39 ( 2008 ) 1555 , 0804.0974 .
phenix , r.a .
lacey , j. phys .
g35 ( 2008 ) 104139 , 0805.1352 .
p. chung and p. danielewicz , phys .
71 ( 2008 ) 1552 , 0807.4892 .
a. kisiel et al . , (
2008 ) , 0808.3363 .
na49 , c. alt et al . , ( 2008 ) , 0809.1445 .
b. tomasik , comput .
commun . 180
( 2009 ) 1642 , 0806.4770 .
p. bozek , phys .
c79 ( 2009 ) 054901 , 0811.1918 .
phobos , b. alver et al .
, ( 2008 ) , 0812.1172 .
w. broniowski et al .
c80 ( 2009 ) 034902 , 0812.3393 .
w. broniowski et al .
, acta phys .
b40 ( 2009 ) 979 , 0812.4935 .
w. florkowski et al .
, acta phys .
b40 ( 2009 ) 1093 , 0901.1251 .
z. chajecki , acta phys .
b40 ( 2009 ) 1119 , 0901.4078 .
p. bozek and i. wyskiel , phys .
c79 ( 2009 ) 044916 , 0902.4121 .
x. luo et al . , phys .
b673 ( 2009 ) 268 , 0903.0024 .
p. bozek and i. wyskiel , phys .
c79 ( 2009 ) 044916 .
j. noronha - hostler et al . , ( 2009 ) , 0906.3960 .
w. broniowski , m. chojnacki and l. obara , phys . rev .
c80 ( 2009 ) 051902 , 0907.3216 .
w. florkowski et al .
phys . a830 ( 2009 ) 821c , 0907.3592 .
p. bozek and i. wyskiel , ( 2009 ) , 0909.2354 .
m. hauer and s. wheaton , phys .
c80 ( 2009 ) 054915 , 0909.2431 .
r. ryblewski and w. florkowski , ( 2009 ) , 0910.0985 .
w. broniowski et al . , ( 2009 ) , 0910.3585 .
brahms , f. videbaek , nucl .
a830 ( 2009 ) 43c , 0907.4742 .
p. bozek , ( 2009 ) , 0911.2392 .
a. kisiel and d.a .
brown , ( 2009 ) , 0901.3527 .
a. kisiel , ( 2009 ) , 0909.5349 .
a. kisiel and t.j .
humanic , ( 2009 ) , 0908.3830 .
w. florkowski and r. ryblewski , ( 2009 ) , 0912.3451 .
heinz , ( 2009 ) , 0901.4355 .
p. bozek and i. wyskiel , ( 2009 ) , 0903.3129 .
lokhtin et al .
, ( 2009 ) , 0910.5129 .
j. noronha - hostler , j. noronha and c. greiner , j. phys .
g37 ( 2010 ) 094062 , 1001.2610 .
luo et al .
, j. phys .
g37 ( 2010 ) 094061 , 1001.2847 .
p. bozek and i. wyskiel , phys .
c81 ( 2010 ) 054902 , 1002.4999 .
star , m.m .
aggarwal et al .
( 2010 ) 022302 , 1004.4959 .
m. hauer , ( 2010 ) , 1008.1990 .
j. noronha - hostler and c. greiner , ( 2010 ) , 1008.5075 .
p. bozek and i. wyskiel - piekarska , ( 2010 ) , 1009.0701 .
a. bilandzic , r. snellings and s. voloshin , ( 2010 ) , 1010.0233 . c. lizhu et al . , ( 2010 ) , 1011.0712 .
p. bozek , w. broniowski and j. moreira , ( 2010 ) , 1011.3354 .
r. brun and f. rademakers , nucl .
instrum . meth . a389 ( 1997 ) 81 , ( http://root.cern.ch ) .
w. florkowski , w. broniowski and m. michalec , acta phys .
b33 ( 2002 ) 761 , nucl - th/0106009 .
w. broniowski and w. florkowski , phys .
87 ( 2001 ) 272302 , nucl - th/0106050 .
p. danielewicz and q.b .
pan , msucl-848 .
e. schnedermann , j. sollfrank and u.w .
heinz , phys .
c48 ( 1993 ) 2462 , nucl - th/9307020 .
f. retiere and m.a .
lisa , phys .
c70 ( 2004 ) 044907 , nucl - th/0312024 .
d. teaney , j. lauret and e.v .
shuryak , ( 2001 ) , nucl - th/0110037 .
t. hirano and k. tsuda , phys .
c66 ( 2002 ) 054905 , nucl - th/0205043 .
p.f . kolb and u. heinz , ( 2003 ) , in quark - gluon plasma 3 , edited by r.c .
hwa and x .-
wang ( world scientific , singapore , 2004 ) , p. 634 , nucl - th/0305084 .
p. huovinen , ( 2003 ) , in quark - gluon plasma 3 , edited by r.c .
hwa and x .-
wang ( world scientific , singapore , 2004 ) , p. 600
, nucl - th/0305064 .
shuryak , nucl .
a750 ( 2005 ) 64 , hep - ph/0405066 .
eskola et al .
c72 ( 2005 ) 044904 , hep - ph/0506049 .
y. hama et al .
a774 ( 2006 ) 169 , hep - ph/0510096 .
t. hirano et al .
b636 ( 2006 ) 299 , nucl - th/0511046 .
p. huovinen and p.v .
ruuskanen , ann .
nucl . part .
56 ( 2006 ) 163 , nucl - th/0605008 .
t. hirano et al .
, j. phys .
g34 ( 2007 ) s879 , nucl - th/0701075 . c. nonaka and s.a .
bass , phys .
c75 ( 2007 ) 014902 , nucl - th/0607018 .
p. huovinen and p. petreczky , ( 2009 ) , 0912.2541 .
g. torrieri et al .
commun . 167
( 2005 ) 229 , nucl - th/0404083 .
g. torrieri et al .
175 ( 2006 ) 635 , nucl - th/0603026 .
s. wheaton and j. cleymans , comput .
commun . 180
( 2009 ) 84 , hep - ph/0407174 .
lokhtin et al .
180 ( 2009 ) 779 , 0809.2708 .
lokhtin et al . , (
2009 ) , 0903.0525 .
amelin et al .
c74 ( 2006 ) 064901 , nucl - th/0608057 .
amelin et al .
c77 ( 2008 ) 014903 , 0711.0835 .
p. koch and j. rafelski , south afr .
j. phys . 9
( 1986 ) 8 .
j. cleymans and h. satz , z. phys .
c57 ( 1993 ) 135 , hep - ph/9207204 .
j. sollfrank et al .
, z. phys .
c61 ( 1994 ) 659 .
p. braun - munzinger et al .
b344 ( 1995 ) 43 , nucl - th/9410026 .
p. braun - munzinger et al .
, phys . lett .
b365 ( 1996 ) 1 , nucl - th/9508020 .
t. csorgo and b. lorstad , phys .
c54 ( 1996 ) 1390 , hep - ph/9509213 .
j. cleymans et al . , z. phys .
c74 ( 1997 ) 319 , nucl - th/9603004 .
j. rafelski , j. letessier and a. tounsi , acta phys .
b27 ( 1996 ) 1037 , nucl - th/0209080 .
j. rafelski , j. letessier and a. tounsi , acta phys .
polon . b28 ( 1997 ) 2841 , hep - ph/9710340 .
f. becattini , j. phys .
g23 ( 1997 ) 1933 , hep - ph/9708248 .
g.d . yen and m.i .
gorenstein , phys .
c59 ( 1999 ) 2788 , nucl - th/9808012 .
j. cleymans and k. redlich , phys .
81 ( 1998 ) 5284 , nucl - th/9808030 . m. gazdzicki and m.i .
gorenstein , acta phys .
b30 ( 1999 ) 2705 , hep - ph/9803462 .
m. gazdzicki , nucl .
a681 ( 2001 ) 153 , hep - ph/9910363 .
p. braun - munzinger , i. heppe and j. stachel , phys .
b465 ( 1999 ) 15 , nucl - th/9903010 .
j. cleymans and k. redlich , phys .
c60 ( 1999 ) 054908 , nucl - th/9903063 .
f. becattini et al .
c64 ( 2001 ) 024901 , hep - ph/0002267 .
p. braun - munzinger et al .
b518 ( 2001 ) 41 , hep - ph/0105229 .
w. broniowski , a. and w. florkowski , acta phys .
b33 ( 2002 ) 4235 , hep - ph/0209286 .
w. florkowski , w. broniowski and p. bozek , j. phys .
g30 ( 2004 ) s1321 , nucl - th/0403038 .
p. bozek , w. broniowski and w. florkowski , acta phys . hung .
a22 ( 2005 ) 149 , nucl - th/0310062 .
w. broniowski , a. baran and w. florkowski , aip conf .
proc . 660 ( 2003 ) 185 , nucl - th/0212053 .
w. florkowski , w. broniowski and a. baran , ( 2004 ) , nucl - th/0412077 .
particle data group , c. amsler et al .
b667 ( 2008 ) 1 .
r. hagedorn , nuovo cim
. suppl . 3 ( 1965 ) 147 .
r. hagedorn and j. ranft , nuovo cim .
( 1968 ) 169 .
r. hagedorn , invited talk at nato advanced study workshop on hot hadronic matter : theory and experiment , divonne - les - bains , france , 27 jun - 1 jul 1994 .
w. broniowski and w. florkowski , phys .
b490 ( 2000 ) 223 , hep - ph/0004104 .
w. broniowski , ( 2000 ) , hep - ph/0008112 .
w. broniowski , w. florkowski and l.y .
glozman , phys .
d70 ( 2004 ) 117503 , hep - ph/0407290 .
f. cooper and g. frye , phys .
d10 ( 1974 ) 186 .
misner , k.s .
thorne and j.a .
wheeler , .
g. bertsch , m. gong and m. tohyama , phys .
c37 ( 1988 ) 1896 .
s. pratt , phys .
d33 ( 1986 ) 1314 .
bugaev , nucl .
a606 ( 1996 ) 559 , nucl - th/9906047 .
k.a . bugaev and m.i .
gorenstein , ( 1999 ) , nucl - th/9903072 .
bugaev , phys .
90 ( 2003 ) 252301 , nucl - th/0210087 .
g. colangelo , j. gasser and h. leutwyler , nucl .
b603 ( 2001 ) 125 , hep - ph/0103088 .
the alice , k. aamodt et al . , ( 2010 ) , 1011.3914 .
p. bozek , ( 2010 ) , 1012.5927 .
p. bozek , ( 2011 ) , 1101.1791 .
j. letessier and j. rafelski , phys .
c59 ( 1999 ) 947 , hep - ph/9806386 .
a. muronga , phys .
88 ( 2002 ) 062302 , nucl - th/0104064 .
d. teaney , phys .
c68 ( 2003 ) 034913 , nucl - th/0301099 .
r. baier , p. romatschke and u.a .
wiedemann , phys .
c73 ( 2006 ) 064903 , hep - ph/0602249 .
r. baier and p. romatschke , eur .
j. c51 ( 2007 ) 677 , nucl - th/0610108 .
p. romatschke and u. romatschke , phys .
99 ( 2007 ) 172301 , 0706.1522 .
chaudhuri , phys .
c74 ( 2006 ) 044904 , nucl - th/0604014 .
u.w . heinz , h. song and a.k .
chaudhuri , phys .
c73 ( 2006 ) 034904 , nucl - th/0510014 .
p. bozek , acta phys .
b39 ( 2008 ) 1375 , 0711.2889 .
h. song and u.w .
heinz , phys .
c77 ( 2008 ) 064901 , 0712.3715 .
p. bozek , acta phys .
b39 ( 2008 ) 1539 , 0803.4447 . h. song and u.w .
heinz , ( 2008 ) , 0812.4274 .
m. luzum and p. romatschke , phys .
c78 ( 2008 ) 034915 , 0804.4015 .
p. bozek , acta phys .
b40 ( 2009 ) 987 , 0901.2272 .
p. bozek , ( 2009 ) , 0911.2397 .
p. romatschke , ( 2009 ) , 0902.3663 .
brown and p. danielewicz , phys .
b398 ( 1997 ) 252 , nucl - th/9701010 .
particle data group , k. hagiwara et al .
d66 ( 2002 ) 010001 . | we present an extended version of therminator , a monte carlo event generator dedicated to studies of the statistical production of particles in relativistic heavy - ion collisions .
the increased functionality of the code contains the following features : the input of any shape of the freeze - out hypersurface and the expansion velocity field , including the 3 + 1 dimensional profiles , in particular those generated externally with various hydrodynamic codes .
the hypersufraces may have variable thermal parameters , which allows for studies departing significantly from the mid - rapidity region , where the baryon chemical potential becomes large .
we include a library of standard sets of hypersurfaces and velocity profiles describing the rhic au+au data at @xmath0 gev for various centralities , as well as those anticipated for the lhc pb+pb collisions at @xmath1 tev . a separate code , femto - therminator , is provided to carry out the analysis of femtoscopic correlations which are an important source of information concerning the size and expansion of the system .
we also include several useful scripts that carry out auxiliary tasks , such as obtaining an estimate of the number of elastic collisions after the freeze - out , counting of particles flowing back into the fireball and violating causality ( typically very few ) , or visualizing various results : the particle @xmath2-spectra , the elliptic flow coefficients , and the hbt correlation radii .
we also investigate the problem of the back - flow of particles into the hydrodynamic region , as well as estimate the elastic rescattering in terms of trajectory crossings .
the package is written in c++ and uses the cern root environment . ,
, , , relativistic heavy - ion collisions , statistical hadronization , monte carlo event generator , lhc , rhic , sps , fair , nica * program summary * _ title of the program : _
+ _ catalog identifier : _ + _ program web page : _ + http://therminator2.ifj.edu.pl/ + _ licensing provisions : _ none + _ computer : _ any computer with a c++ compiler and the cern root environment , ver . 5.26 or later , tested with intel core2 duo cpu e8400 @ 3 ghz , 4 gb ram + _ operating system under which the program has been tested : _ + linux ubuntu 10.10 x64 ( gcc 4.4.5 ) root ver . 5.26
+ _ programming language used : _ with the cern root libraries , bash shell + _ no . of lines in distributed program , including test data : _
+ 13000 code , 300 configuration , 400000 input files ( hypersurface library ) + _ no . of bytes in distributed program ,
including test data : _ 2.7 mb + _ distribution format : _ tar.gz
+ _ nature of physical problem : _ particle production via statistical hadronization in relativistic heavy - ion collisions + _ method of solution : _ monte - carlo simulation , analyzed with root + _ restrictions concerning the complexity of the problem : _ none + _ memory required to execute with typical data : _ + 30 mb therm2_events , + 150 mb therm2_femto + _ typical running time : _ default configuration at 3 ghz + primordial multiplicities 70 min ( calculated only once per case ) , + 8 min/500 events , + 10 min draw all figures , + 25 min / one @xmath3 bin in the hbt analysis with 5000 events + _ typical data file size : _ default configuration + 45 mb/500 events + 35 mb / correlation file ( one @xmath3 bin ) + 45 kb / fit file ( projections and fits ) + |
in a series of intermetallic compound rmn@xmath0 ( r = rare earth ) , an existence of magnetic moments on mn sites depends largely on an inter - atomic mn - mn distance@xcite . above a critical distance of @xmath11=2.7
@xmath12 m there exists a magnetic moment on the mn which is antiferromagnetically ordered in general . below this critical distance
the ground state remains paramagnetic down to lowest temperature .
this fact that the mn - mn inter - atomic distance plays a key - role in determining their magnetic properties indicates that these compounds may be magnetically sensitive to such a perturbation as a chemical substitution or an application of pressure .
an interesting possibility is then that an application of pressure induces quantum critical phase transition governed by the zero point quantum spin fluctuation . in particular
this is the case of ymn@xmath0 which exhibits antiferromagnetic ( af ) order with a nel temperature around 100 k at ambient pressure .
a relatively low pressure of @xmath13=3 to 4 kbar is enough to suppress the af order@xcite .
the paramagnetic to af magnetic transition upon cooling is of first order type accompanying by well pronounced hysteretic phenomena in various measurements .
therefore , the af to paramagnetic phase transition across @xmath14 is anticipated to be first order type .
so far it is reported that the paramagnetic state at ambient pressure is dominated by the af spin fluctuations @xcite which is well treated in terms of the self - consistent renormalized ( scr ) spin fluctuation theory developed by moriya et al . @xcite . as a matter of fact
, the sc - substitution of y suppresses the af order and the spin fluctuations in the paramagnetic phase was reported to be well described in terms of the scr theory @xcite . on the other hand , the inelastic neutron scattering experiment has found that the singlet correlation among four mn spins forming the tetrahedron is developed to stabilize this paramagnetic phase , i.e. , a kind of spin liquid state is formed in the sc substituted system .
thus , it is still controversial whether or not this novel character of spin fluctuations is actually related to the nature in the paramagnetic state above @xmath1 in ymn@xmath0 because the effect of disorder induces by the sc -substitution is not addressed yet . in other words ,
we address the question whether or not , near the critical point in ymn@xmath0 where the nel temperature is suppressed to zero , the af spin fluctuations are also critical with a damping energy tending to zero .
an application of pressure to suppress the af order is promising to settle these issues , because this is the only way to stabilize the paramagnetic state in ymn@xmath0 at low temperatures without any disorderliness .
the aim of this paper is to enligthten the evolution of low - energy magnetic excitations in the pressure - induced paramagnetic phase by approaching the critical point from the higher pressure side .
a polycrystalline sample has been crushed in fine powder in order to ensure the penetration of the radio - frequency pulsed field .
a conventional clamp - type pressure cell was employed for the nmr experiment under pressure with use of the fluorinert liquid as the pressure transmitting medium .
the fluorinert volume was of the proportion as the sample volume and froze below 200 k. to calibrate the pressure at the sample position at low temperature , the pressure shift in @xmath15 of lead powder was monitored .
the measurement of @xmath15 was performed by an inductive method using the in - situ nmr coil in a resonant circuit .
the nqr spectrum has been obtained by plotting the spin - echo intensity as a function of frequency with an interval of 50 khz .
the spin - lattice relaxation time , @xmath16 , was measured by the saturation recovery method in a temperature range of @xmath17 k. for the nqr transition of @xmath18 , the recovery function of nuclear magnetization , m , is given by @xmath19 and the spin - lattice relaxation time was uniquely determined by a fit of the data with the above formula .
fig . [ fig1 ] indicates the evolution of the @xmath20mn nqr spectrum for the @xmath21 transition at @xmath22 k under pressure . with increasing pressure
, the spectrum experiences a slight shift towards a higher frequency side , reflecting an increase of the electric field gradient ( efg ) .
the full - width at half - maximum in a series of nqr spectra does not change appreciably , which means that the pressure - induced distribution in efg can be ignored .
the nuclear spin - lattice relaxation time @xmath16 was not measured above 120 k in order to avoid a melting of the transmitting liquid medium at high temperatures which increases the pressure effectively . the @xmath23 dependence of @xmath6 , which is plotted with the semi - logarithmic scale in temperature ( fig .
[ fig2 ] ) , shows the possible existence of two regimes marked by a kink ( arrows ) .
hereafter , we will refer to this cross - over temperature as a characteristic spin fluctuation temperature , @xmath24 , in the pressure - induced paramagnetic phase .
@xmath5 increases from 30 k with increasing pressure . below @xmath5 , @xmath6 becomes weakly temperature dependent .
the spin - lattice relaxation rate , @xmath25 , is in general related to the transverse component of the imaginary part in the dynamical magnetic susceptibility @xmath26 such as @xmath27 where @xmath28 is the gyromagnetic ratio of the nuclei studied , @xmath29 the number of atoms , @xmath30 the hyperfine coupling constant , and @xmath31 the nmr frequency ( @xmath32 represents the transverse component of the generalized susceptibility ) . in the case of nearly antiferromagnetic metals , the q dependence of the low - energy spin fluctuations
are dominated by the components close of the af vector * * q**=*q * and the main contribution will thus be such as @xmath33 . therefore eq . ( [ equ2 ] ) can be given by a simple form of @xmath34 the pressure and thermal dependences of @xmath6 thus allow us to probe corresponding dependences of af spin fluctuation . in the framework of the scr theory the dynamical susceptibility is given by @xmath35 where @xmath36 is the parameter to make the self - consistency equation valid , the non - interacting susceptibility @xmath37 and @xmath38 the coulomb intra - site interaction .
the denominator of eq .
( [ equ4 ] ) is equivalent to the stoner factor which tends to zero at the criticality and @xmath39 .
correspondingly the divergence of @xmath40 means the divergence in the static staggered susceptibility and as a result , it is expected that @xmath41 undergoes a divergence at @xmath39 or it is largely enhanced approaching @xmath42 .
however , the experiment does not follow the above expectation .
rather , the development of af spin fluctuations is saturated upon cooling and approaching the @xmath39 .
namely , @xmath5 does not tend to zero , but seems to remain finite even at @xmath39 .
this may due to the first order nature of the transition from the af state to paramagnetic at @xmath39 induced by pressure . above
@xmath5 , the spin fluctuations are dominated by thermal fluctuations which are treated in terms of the scr theory , whereas 1/@xmath43 approaches a constant value below @xmath5 characteristic for the fermi liquid excitation .
[ fig3 ] indicates the pressure dependence of @xmath44 deduced from the present @xmath16 measurement . from the electrical resistivity measurements under pressure , the spin fluctuation temperature was defined as the cross - over temperature below which the thermal variation of @xmath45 obeys the expected fermi liquid @xmath46 law@xcite
remarkably , a fairly good agreement is obtained concerning the variation of both @xmath47 under pressure @xmath48 with these experimental implications , it is possible that the development in spin fluctuation remains in short - range upon cooling .
from the thermal variation of the nuclear spin lattice relaxation rate , @xmath49 of @xmath20mn above the critical pressure of 4 kbar , the spin fluctuation feature has been found to change below the temperature @xmath5 . at
temperatures higher than @xmath5 , the nuclear relaxation behavior is well described in terms of the self - consistent renormalized spin fluctuation theory for a nearly af metal , whereas below @xmath5 , a deviation from that is significant and @xmath50 exhibits a weak temperature variation . it is pointed out that @xmath47 coincides with the temperature below which a @xmath46 law in resistivity is valid as expected for the fermi liquid state .
we proposed that these features are understood from the view that the development in af spin fluctuation remains in short - range associated with the singlet formation among mn spins in each tetrahedron as suggested by the inelastic neutron experiments on y@xmath8sc@xmath9mn@xmath0 . | we report the nqr study on the pressure - induced paramagnetic state in the antiferromagnetic ( af ) intermetallic compound ymn@xmath0 with the nel temperature @xmath1=100 k at ambient pressure . from the @xmath2 variation of the nuclear spin - lattice relaxation rate , @xmath3 of @xmath4mn above the critical pressure of 4 kbar
, the spin fluctuation feature is found to change below a temperature @xmath5 . at higher temperature than @xmath5
, the nuclear relaxation behavior is well described in terms of the self - consistent renomalized ( scr ) spin fluctuation theory for nearly af metals , whereas below @xmath5 a deviation is significant and @xmath6 exhibits a weak @xmath2 variation .
it is pointed out that @xmath5 coincides with the temperature below which a @xmath7 law in electrical resistivity is valid as expected for a fermi liquid ground state .
we proposed that these features are understood from the standpoint that the development of af spin fluctuation remains in short - range associated with the singlet formation among mn spins in each tetrahedron as suggested by the inelastic neutron experiments on y@xmath8sc@xmath9mn@xmath0 .
# 1@xmath10#1 pnumwidth2em |
the quantitative description of real - time dynamics in strongly correlated systems is one of the outstanding challenges of contemporary condensed - matter physics , with relevance to varied systems ranging from cold atoms @xcite and dissipative systems @xcite to quantum - dot devices @xcite and biological donor - acceptor molecules .
@xcite alongside fundamental question concerning the underlying time scales and the long - time behavior , there are many technological issues that require careful investigation . for example , the decoherence and relaxation of spins appears to be the major obstacle for the realization of quantum - computing algorithms in real systems .
@xcite another key issue is the understanding of coherent control and the switching characteristics of nanodevices such as single - electron transistors .
@xcite these and related topics require the development and application of suitable many - body techniques .
@xcite over the years , the kadanoff - baym @xcite and keldysh @xcite techniques have proven to be accurate tools for describing the real - time dynamics of weakly correlated systems such as light - matter interaction in semiconductors @xcite and the decoherence and relaxation of an impurity spin well above the kondo temperature .
@xcite geared toward perturbation theory , these techniques generally fail upon the development of strong correlations , when nonperturbative approaches are in order . a case in point
are quantum dots tuned to the kondo regime,@xcite where traditional diagrammatic - based approximations are unsuitable to describe the nonequilibrium state.@xcite the difficulty lies in the fact that strongly correlated systems change their nature as a function of certain control parameters such as the temperature or the coupling constants , an aspect well captured by renormalization - group approaches.@xcite the precise status of a voltage bias as yet another control parameter in interacting nanostructures is still under debate .
recent years have witnessed an impressive advancement of numerical techniques aimed at tracking the real - time dynamics of strongly correlated systems , primarily with the development of the time - dependent density - matrix renormalization group ( td - dmrg ) .
@xcite yet while the adaptive td - dmrg works remarkably well on time scales of order the reciprocal bandwidth , it is presently unsuited for tackling longer time scales due to an accumulated error that grows first linearly and then exponentially with the time elapsed .
alternative formulations @xcite of the td - dmrg circumvent the accumulated error , but are simply too demanding to advance to long times .
recent adaptations of continuous - time monte carlo techniques to real - time dynamics @xcite are free of finite - size effects , but are confined to short time scales due to an inherent sign problem .
the chebyshev expansion technique , @xcite developed by tal ezer and kosloff , @xcite offers yet another extremely powerful approach for tracking the time evolution of finite - size systems .
however , it too is quite limited in accessing long time scales in the presence of interactions due to the exceedingly large hilbert space that must be retained .
a complementary approach is provided by the time - dependent numerical renormalization group ( td - nrg ) .
@xcite the td - nrg can successfully bridge over vastly different time scales , but is far more restrictive in the systems and perturbations to which it can be applied . a composite approach that combines the complementary traits of the different techniques mentioned above
is highly desirable . in this paper
, we devise such a hybrid approach that combines the outstanding capabilities of the numerical renormalization group ( nrg ) to systematically construct the effective low - energy hamiltonian of the system @xcite with the prowess of complementary approaches for calculating the real - time dynamics derived from this hamiltonian .
typically strongly correlated systems possess multiple energy scales that markedly differ in magnitude , hence their dynamics is governed by vastly different time scales . this spread of time scales , which may differ by many orders of magnitude , poses an enormous obstacle for most computational approaches .
the td - nrg is quite unique in this respect as it can efficiently bridge between the different time scales .
our hybrid approach presented below provides a flexible platform for combining the td - nrg with one s method of choice for treating the effective low - energy hamiltonian .
possible choices for the complementary method could be exact diagonalization , the td - dmrg , and possibly also the time - dependent noncrossing approximation .
@xcite here we shall demonstrate the applicability of our approach by combining the td - nrg with the chebyshev expansion technique ( cet ) , providing thereby an important proof of principle .
the basic philosophy underlying the hybrid - nrg is to first exhaust the td - nrg in order to decompose the time - dependent wave function into distinct components , each associated with a separate time scale and evolving according to its own reduced hamiltonian acting on a suitable subspace of the full fock space .
each of the individual components with its associated hamiltonian can then be treated with improved accuracy using , e.g. , the td - dmrg or cet . in this manner
one can exploit the successive reduction in energy scales in order to boost the td - dmrg and cet to long time scales that otherwise would be inaccessible to either of these methods .
concomitantly , the accuracy and flexibility of the td - nrg are greatly enhanced , as we demonstrate by extending the approach to the physically relevant case of repeated switchings .
the hybrid platform further offers an appealing way to reduce discretization errors inherent to the wilson chain by converting to hybrid chains .
the idea to use the nrg level flow to construct effective hamiltonians is , of course , not new .
dating back to the original work of wilson , @xcite this framework has been significantly advanced by hewson @xcite who used it as a starting point for devising a renormalized perturbation theory .
hewson s approach requires , however , analytical knowledge of the form of the low - energy hamiltonian and its associated quantum field theory .
so far it has been applied mainly to the single - impurity anderson model , where it was used , among other things , to calculate the steady - state current @xcite in the limit of a small bias voltage .
our approach is far more general as it makes no assumption on the analytical form of the effective hamiltonian .
rather , it is solemnly based on wilson s original concept @xcite that the nrg level flow contains all accessible information , and in particular can accurately describe the crossover region between two distinct fixed points ( a regime which generally lies outside the reach of perturbative methods ) .
our framework exclusively uses the sequence of diagonalized nrg hamiltonians , @xcite circumventing thereby any prejudice on the form of the effective hamiltonian . as a result
our method is model independent , relying solely on the nrg approach itself . in this paper
, we extend the original td - nrg algorithm from a single quantum quench to multiple switchings , which requires an additional approximation beyond the one underlying the conventional td - nrg . as we demonstrate by explicit calculations ,
the quality of the approximation can be systematically improved by enlarging the subspace treated using the complementary method . in the td - nrg
, the continuous bath is represented by a discretized wilson chain , @xcite characterized by tight - binding hopping matrix elements that decay exponentially along the chain .
this separation of energy scales enables access to exponentially long time scales @xcite that otherwise would be inaccessible using ordinary tight - binding chains . in analyzing the accuracy of the td - nrg
it is important to distinguish between two sources of error : one extrinsic due to the discretized representation of the continuous bath in terms of a wilson chain , and the other intrinsic due to the td - nrg algorithm for tracking the real - time dynamics on the wilson chain . in the resonant - level model following a sudden quench of the level energy from @xmath0 to @xmath1 at time @xmath2 .
the red solid line shows the exact time evolution on the wilson chain , obtained by exact diagonalization of the single - particle eigenmodes .
the dashed blue line depicts the td - nrg result , while the solid black line shows the exact analytical solution for a continuous band [ given by eq .
( 50 ) of ref . ] .
( b ) a two - dimensional contour plot of the exact time - dependent occupancies @xmath3 of the first 26 sites along the wilson chain , labeled @xmath4 .
the red arrows indicate instances in time when reflected currents reach the impurity site .
the very same times are marked by the black arrows in panel ( a ) .
parameters : @xmath5 , @xmath6 , the number of states kept in the td - nrg is @xmath7 , and the chain length is @xmath8.,scaledwidth=38.5% ] in the resonant - level model following a sudden quench of the level energy from @xmath0 to @xmath1 at time @xmath2 .
the red solid line shows the exact time evolution on the wilson chain , obtained by exact diagonalization of the single - particle eigenmodes .
the dashed blue line depicts the td - nrg result , while the solid black line shows the exact analytical solution for a continuous band [ given by eq .
( 50 ) of ref . ] .
( b ) a two - dimensional contour plot of the exact time - dependent occupancies @xmath3 of the first 26 sites along the wilson chain , labeled @xmath4 .
the red arrows indicate instances in time when reflected currents reach the impurity site .
the very same times are marked by the black arrows in panel ( a ) .
parameters : @xmath5 , @xmath6 , the number of states kept in the td - nrg is @xmath7 , and the chain length is @xmath8.,scaledwidth=46.0% ] as already noted in ref . , the latter source of error is remarkably small .
we illustrate this point in fig .
[ fig : rlm - n - vs - t](a ) for the noninteracting resonant - level model ( rlm ) , describing a single fermionic level coupled by hybridization to a conduction band ( see sec .
[ sec : quench - rlm - hybrid - chain ] for an explicit definition of the model ) . abruptly shifting the energy of the level and tracking the time evolution of the level occupancy @xmath9
, we compare the td - nrg to an exact analytical solution for a continuous bath [ given by eq .
( 50 ) of ref . ] , as well as to an exact numerical solution on the wilson chain using exact diagonalization of the single - particle eigenmodes .
only minuscule deviations are found between the td - nrg and the exact solution on the wilson chain , both of which significantly depart at some point from the continuum - limit result . as analyzed in ref .
, one can decrease the deviations from the continuum limit and delay them to a later time by reducing the wilson discretization parameter @xmath10 . at the same time , the deviations are hardly affected by prolonging the chain length @xmath11 .
this leads to two important conclusions : ( i ) the main source of error in the td - nrg is extrinsic rather than intrinsic ; @xcite ( ii ) the exponentially decaying tight - binding matrix elements , which lie at the heart of the nrg , @xcite are also the limiting factor for reproducing the continuum - limit result for the quench dynamics . to understand the source of the deviations from the continuum - limit result , we note that globally conserved quantities such as the charge or the spin of the system are locally connected by the continuity equation to associated charge and spin currents .
the exponentially decreasing tight - binding matrix elements along the wilson chain significantly slow down the propagation of such currents , @xcite generating internal reflections at the sites of the one - dimensional chain .
this is depicted in fig .
[ fig : rlm - n - vs - t](b ) , where we plot the exact time - dependent occupancies of the first 26 sites along the chain in response to a sudden quench of the impurity level .
the two - dimensional contour plot clearly reveals reflections of the charge current at certain positions and certain characteristic times .
once the reflected currents reach the impurity site its level occupancy starts to deviate significantly from the exact continuum - limit result .
indeed , the red arrows in fig .
[ fig : rlm - n - vs - t](b ) indicate instances in time when reflected charge wave fronts reach the impurity site . at these very same times
the occupancy on the impurity level develops new structures [ marked by the black arrows in fig .
[ fig : rlm - n - vs - t](a ) ] that are absent in the continuum limit . upon decreasing the wilson discretization parameter
@xmath10 the magnitude of the reflected currents is suppressed and the reflection points are pushed deeper down the chain , however the effect is never fully eliminated as long as @xmath12 .
, @xmath6 , and @xmath13 .
the chain length is @xmath8 .
panel ( a ) depicts positive energies on a logarithmic scale .
panel ( b ) shows the full spectrum on a linear scale.,width=264 ] an alternative perspective on the effect of the wilson discretization procedure is provided by examining the exact single - particle energy levels of the rlm . for an ordinary tight - binding chain of length @xmath11 with a constant hopping matrix element @xmath14 , the single - particle energy levels are roughly uniformly distributed in the energy range @xmath15 $ ] , where @xmath16 is the conduction - electron bandwidth . as the chain length
is increased the energy levels become more densely distributed until a continuous spectrum is recovered for @xmath17 .
a different picture applies to the wilson chain . as depicted in fig .
[ fig : rlm - single - particle - levels ] , the single - particle energy levels are uniformly distributed on a logarithmic scale , resulting in a sparse distribution of levels at higher energies . enumerating the positive single - particle energy levels from high to low , these scale as @xmath18 , in accordance with wilson s logarithmic discretization of the conduction band .
@xcite by prolonging the chain length @xmath11 one increases the total number of levels , yet the distribution of high - energy levels retains its sparse form even as @xmath17 .
a continuous band is recovered only upon implementing the combined limit @xmath19 , @xmath17 , which illustrates the limitation of working with a fixed @xmath12 .
these argumentations clearly point to an intrinsic tradeoff within the td - nrg , as the very same logarithmic discretization that enables access to exponentially long time scales also prevents from fully recovering the continuum limit .
the question then arises whether one can somehow reconcile these two apparently contradicting properties , which is one of the main goals of this work
. briefly stated , the aim of this paper is a two - fold extension of the td - nrg .
the first goal is to devise a flexible framework for hybridizing the td - nrg with complementary methods of calculating real - time dynamics that do not rely on the special structure of the wilson chain . by liberating ourselves from working on the wilson chain at all time
scales , the hybrid method should enable a systematic improvement of both finite - size and discretization errors .
the second objective is to extend the method from a single quantum quench to repeated switchings between two distinct hamiltonians @xmath20 and @xmath21 . besides being of considerable interest on its own right
, this protocol can be viewed as a first vital step toward treating a general time - dependent hamiltonian @xmath22 . indeed , discretizing the time axis and replacing @xmath22 within each time interval @xmath23 with the constant form @xmath24
, the full time evolution can be interpreted as a sequence of quenches . despite its conceptual simplicity
this strategy has not been pursued thus far since the initial nonequilibrium density operator @xmath25 at the beginning of each time interval is neither explicitly known nor of the analytical form required by the original td - nrg formulation .
@xcite to this end , we derive the hybrid - nrg in sec .
[ sec : hybrid - nrg - theory ] .
since the formulation relies heavily on the td - nrg , we shall commence in sec .
[ sec : td - nrg ] with a brief review of the td - nrg following ref . .
this exposition is essential not only for keeping the paper self - contained , but mainly for introducing the notations and tools that will be used throughout our construction of the hybrid - nrg . after setting the stage in sec .
[ sec : td - nrg ] , we present the first major conceptual result of this paper in sec .
[ sec : hybrid - tdnrg ] : the hybrid - nrg approach .
the key idea is to partition the wilson chain into two parts : the high - energy part is treated with the td - nrg while the low - energy part is feed into the complementary method of choice .
the standard td - nrg approach can be embedded into this more general hybrid framework by shifting the partition to the end of the wilson chain . in order to make contact with the complementary approach used for hybridizing with the td - nrg ( e. g. , the cet )
, we examine the interface with the td - nrg from a wave - function perspective in sec . [
sec : expectation - values ] . using the central results of sec .
[ sec : hybrid - nrg - theory ] , we extend the hybrid approach to periodic switching between two hamiltonians in sec . [ sec : switchings ] , making the simulation of coherent control by an external field accessible to the hybrid method .
this section covers the second major conceptual result of this paper .
since the exact evaluation of the reduced density matrix of an arbitrary density operator has remained a computational challenge , we propose a set of approximations which neglect some of the high - energy contributions to the real - time dynamics .
these approximations are systematic and are carefully analyzed .
the complementary method used to supplement the td - nrg in this paper , the chebyshev expansion technique,@xcite is reviewed in sec . [ sec : chebyshev ] .
since the deviations of finite - size nonequilibrium dynamics from the continuum limit are linked to the bath discretization , our hybrid framework targets the liberation of numerical simulations from the particular form of the wilson chain without losing access to exponentially long time scales . in sec .
[ sec : quench - rlm - hybrid - chain ] , we demonstrate the potential of our hybrid framework by investigating the influence of different hybrid chain types on the discretization errors encountered in the local dynamics . in sec .
[ sec : periodic - switching ] , we present our results for periodic switching . since the td - nrg is embedded in our hybrid framework , we first discuss in sec .
[ sec : td - nrg - periodic - switching ] the limitations of a simple extension of the td - nrg to periodic switching
. the exact solution of the finite - size rlm subject to a periodic drive is used to benchmark both the hybrid nrg - cet and the periodic td - nrg .
the interacting resonant - level model @xcite ( irlm ) serves as a first nontrivial test of our hybrid approach .
it includes an additional local capacitive coupling @xmath26 between the charge on the impurity level and the fermionic band , preventing an exact solution of its dynamics .
the low - energy fixed point of the irlm is identical to that of the noninteracting rlm , featuring a renormalized , @xmath26-dependent hybridization strength . by comparing the real - time dynamics of both models with identical renormalized hybridization strengths
, we show in sec .
[ sec : irlm ] that the local charge dynamics significantly deviates with increasing @xmath26 from the noninteracting case .
interaction - induced oscillations are found in the irlm , whose characteristic frequency depends only on the renormalized hybridization and not directly on @xmath26 .
those oscillations are completely absent in the noninteracting rlm , even though both models share the same low - energy fixed point . using a strong - coupling analysis
, we provide a simple physical picture for this surprising effect .
finally , we conclude with a discussion and outlook in sec .
[ sec : discussion - outlook ] .
the td - nrg has been designed to track the real - time dynamics of quantum - impurity systems following an abrupt quantum quench .
the perturbations under consideration are implicitly assumed to be of local character , i.e. , perturbations that are applied either to the impurity itself or to its close vicinity .
the hamiltonian of a quantum - impurity system has the generic structure @xmath27 where @xmath28 models the continuous bath , @xmath29 represents the decoupled impurity , and @xmath30 describes the coupling between the two subsystems .
the entire system is characterized at time @xmath2 by the density operator @xmath31 when a static perturbation @xmath32 is suddenly switched on : @xmath33 .
the density operator evolves thereafter in time according to @xmath34 our objective is to use the nrg to compute the time - dependent expectation value @xmath35 of a general local operator @xmath36 . as shown in ref .
, the result can be written in the form @xmath37 where @xmath38 and @xmath39 are the dimension - full nrg eigenenergies of the perturbed hamiltonian at iteration @xmath40 , @xmath41 is the matrix representation of @xmath36 at that iteration , and @xmath42 is the reduced density matrix defined in eq .
( [ eqn : reduced - dm - def ] ) below .
the restricted sum over @xmath43 and @xmath44 requires that at least one of these states is discarded at iteration @xmath45 .
the nrg chain length @xmath11 implicitly defines the temperature entering eq .
( [ rho-0 ] ) : @xmath46 , where @xmath12 is the wilson discretization parameter . the derivation of eq .
( [ eqn : time - evolution - intro ] ) relies on two key ingredients : ( i ) the identification of a complete basis set of approximate nrg eigenstates for the many - body fock space @xmath47 of the wilson chain ; ( ii ) expectation values are obtained by explicitly tracing over this complete basis set using a suitable resummation procedure .
below we review these two key components following the notations and presentation of ref . .
is divided into a subchain of length @xmath45 and the `` environment '' @xmath48 .
the hamiltonian @xmath49 can be viewed either as acting only on the subchain of length @xmath45 , or as acting on the full chain of length @xmath11 , but with the hopping matrix elements @xmath50 all set to zero .
the former picture is the traditional one . in the td - nrg
one adopts the latter point of view.,scaledwidth=48.0% ] the nrg targets an iterative solution of a quantum impurity coupled to a finite wilson chain with @xmath11 chain links .
bath sites we refer to it either as an @xmath11-site chain or a chain of length @xmath11 , so as to emphasize the number of distinct hopping matrix elements or energy scales involved . ]
similar to the initial sweep in the finite - size dmrg , one can view the nrg procedure as a set of operations , where at first all hopping matrix elements are set to zero along the @xmath11-site chain , and at each successive step another hopping matrix element is switched on .
the full hamiltonian @xmath51 is recovered once all hopping matrix elements have been switched on .
the entire sequence of hamiltonians @xmath49 with @xmath52 act in this picture on the same fock space @xmath47 of the @xmath11-site chain , hence each nrg eigenenergy of @xmath49 has an extra degeneracy of @xmath53 , where @xmath54 is the number of distinct configurations at each site along the chain .
the extra degeneracy stems from the @xmath55 `` environment '' sites at the end of the chain , denoted by @xmath56 in fig . [
fig : semi - infinite - nrg ] , which remain decoupled from the impurity at iteration @xmath45 .
when acting on the @xmath45-site chain , we label the nrg eigenstates and eigenenergies of @xmath49 by @xmath57 and @xmath58 , respectively .
consider now the action of @xmath49 on the full @xmath11-site chain .
enumerating the different configurations of site @xmath23 by @xmath59 , each of the tensor - product states @xmath60 with arbitrary @xmath61 is a degenerate nrg eigenstate of @xmath49 with energy @xmath58 . to label these states we introduce the shorthand notation @xmath62 , where the `` environment '' variable @xmath63 encodes the @xmath64 site labels @xmath61 , and the index @xmath45 is used to record where the chain is partitioned into a `` subsystem '' and an `` environment '' ( see fig .
[ fig : semi - infinite - nrg ] ) . in order to retain a manageable number of states ,
the high - energy eigenstates are discarded after each iteration , which is fully justified in equilibrium by the hierarchy of energy scales along the wilson chain and the boltzmannian form the equilibrium density operator . regarding all states of the final iteration
as discarded , it has been show in refs . that the collection of all states discarded in the course of the nrg iterations form a complete basis set of approximate nrg eigenstates for the full @xmath11-site chain .
to understand this important point , consider the first iteration @xmath65 at which states are discarded . in order to keep track of the complete basis set of the @xmath11-site chain , the eigenstates @xmath66
can be formally divided into two distinct subsets : the discarded high - energy states @xmath67 and the kept low - energy states @xmath68 . obviously , the sum of the two subsets form a complete basis set of the full chain . to simplify the notations
we shall omit hereafter the subscripts @xmath69 and @xmath70 , and will use in exchange the indices @xmath71 and @xmath72 to label the discarded and kept states , respectively . at the next nrg iteration
only the kept states are used to construct the nrg eigenstates of @xmath73 within the truncated subspace spanned by @xmath74 .
the resulting nrg eigenstates can again be divided into two subsets of discarded and kept states which , when combined with the discarded eigenstates of iteration @xmath65 , form a complete basis set of the fock space @xmath47 of the full @xmath11-site chain . repeating this procedure at all subsequent iterations ,
one continues to maintain a complete basis set of @xmath47 up to the final nrg iteration @xmath75 . in this manner
we arrive at the following completeness relation @xmath76 where the summation over @xmath45 starts from the first iteration @xmath65 at which a basis - set reduction is imposed . here
the summation indices @xmath71 and @xmath77 implicitly depend on @xmath45 , and the projector onto the subspace discarded at iteration @xmath78 is defined as @xmath79 the complementary projector @xmath80 onto the subspace retained at iteration @xmath45 ( @xmath81 ) is given in turn by @xmath82 which can be recast in the form @xmath83 this latter equality reflects the fact that all states retained at iteration @xmath45 are necessarily discarded at some later iteration @xmath84 . in particular , since all states of the final iteration @xmath11 are regarded discarded , then @xmath85 is identically zero while @xmath86 coincides with @xmath87
. combined with eq .
( [ eqn : p - kept - states ] ) , the completeness relation of eq .
( [ equ : complete - basis ] ) can further be partitioned into @xmath88 with arbitrary @xmath89 .
this useful identity will be repeatedly used in constructing the hybrid - nrg . following a sudden quench ,
the time evolution of the system is governed by the perturbed hamiltonian @xmath90 while the initial condition is encoded in the initial density matrix @xmath91 . for quantum - impurity systems ,
all relevant information on the initial condition is contained in eq .
( [ eqn : time - evolution - intro ] ) in the form of the reduced density matrices @xcite @xmath42 , defined as @xmath92 here the states @xmath93 and @xmath94 correspond to the hamiltonian @xmath90 , and the summation runs over the environment degrees of freedom @xmath77 .
the only approximation entering eq .
( [ eqn : time - evolution - intro ] ) is the standard nrg approximation @xmath95 , which enables us to write @xmath96 apart from this sole point , eq .
( [ eqn : time - evolution - intro ] ) constitutes an exact evaluation of @xmath35 on the discretized @xmath11-site chain .
practical calculations hinge on the ability to accurately compute the reduced density matrices of eq .
( [ eqn : reduced - dm - def ] ) . for a general @xmath91
this can be a daunting task .
however , in the case of interest where @xmath91 has the standard boltzmann form of eq .
( [ rho-0 ] ) , the summation over @xmath77 can be carried out exactly .
hence @xmath42 can be evaluated at the same level of accuracy as the equilibrium density operator @xmath91 .
technically this goal is achieved by implementing two independent nrg runs , one for the initial hamiltonian @xmath97 in order to construct @xmath91 using to eq .
( [ rho-0 ] ) , and another for the full hamiltonian @xmath90 .
the reduced density matrix @xmath42 is first evaluated with respect to the eigenstates of the initial hamiltonian , and then rotated @xcite to the eigenstates of the full hamiltonian using the overlap matrices @xmath98 here @xmath99 denotes an nrg eigenstate of the full hamiltonian at iteration @xmath45 , and @xmath100 is an nrg eigenstate of the initial hamiltonian at the same iteration .
the method of calculating the overlap matrices @xmath101 is detailed in ref . .
the original td - nrg approach , summarized above , tracks the quench dynamics of a quantum - impurity system in terms of the phase factors @xmath102 and the reduced density matrices @xmath42 assigned to each nrg iteration @xmath45 .
although quite elegant and useful , it is less transparent how to incorporate ideas from methods such as the td - dmrg or cet , as these deal with wave functions directly . to develop a convenient and flexible interface between the td - nrg and these vastly different approaches
we reformulate the former approach from a wave - function perspective .
let us commence with accurately stating the problem from a wave - function perspective .
we are interested in tracking the time evolution of some initial state @xmath103 under the dynamics defined by the hamiltonian @xmath104 acting on a finite wilson chain of length @xmath11 .
we shall not concern ourselves at this stage with how the initial state @xmath103 is generated , but will elaborate on this important point later on .
formally our task boils down to computing @xmath105 application of the completeness relation of eq .
( [ completeness-2 ] ) to the state @xmath106 leads to its partitioning according to @xmath107 where @xmath108 and @xmath109 are the projections of @xmath106 onto the subspaces defined by @xmath110 and @xmath111 , respectively .
equation ( [ partitioning - of - psi ] ) simply converts the general state @xmath106 into a concrete representation in terms of our complete basis set .
given the time - evolved wave function of eq .
( [ partitioning - of - psi ] ) , we proceed to compute time - dependent averages of physical observables : @xmath112 to this end , we use the completeness relation of eq .
( [ completeness-2 ] ) to decompose any arbitrary operator @xmath113 into @xmath114 where the summations over @xmath45 and @xmath84 start from @xmath65 .
writing the first two terms on the right - hand side of eq .
( [ a - via - pm ] ) as @xmath115\end{aligned}\ ] ] and noting that @xmath116 the operator @xmath113 is recast in the exact form @xmath117 where @xmath118 and @xmath119 here the index @xmath120 can take any value in the range @xmath121 .
explicitly , the operator @xmath122 has the formal representation @xmath123 where the restricted sum @xmath124 implies , as before , that at least one of the states @xmath43 and @xmath44 is discarded at iteration @xmath45 . as in the original td - nrg
, we focus hereafter on local operators @xmath113 that act solely on degrees of freedom that reside either on the impurity itself or on the first @xmath65 sites along the wilson chain .
@xcite for any such local operator , the matrix elements in eq .
( [ a - of - t-2 ] ) are diagonal in and independent of the environment variables @xmath77 and @xmath125 : @xmath126 substituting the operator decomposition of eq .
( [ eqn : operator - projection ] ) into eq .
( [ eqn : def - expection - a ] ) and using the definition @xmath127 of eq .
( [ chi_m_t ] ) , the time - dependent expectation value takes the form @xmath128 this general expression reduces for a local operator to @xmath129 where @xmath130 is the reduced density matrix at iteration @xmath45 .
three comments should be made about eqs .
( [ a - of - t-3 ] ) and ( [ rho - m - of - t ] ) .
first , these expressions are both general and exact for the real - time dynamics on the discretized chain .
apart from the restriction to local operators , no further approximations or assumptions are involved .
second , eqs .
( [ a - of - t-3 ] ) and ( [ rho - m - of - t ] ) can be easily extended to a statistical admixture of initial states @xmath131 with the statistical weights @xmath132 .
this requires the simple substitutions @xmath133 and @xmath134 in eqs .
( [ a - of - t-3 ] ) and ( [ rho - m - of - t ] ) , respectively .
third , the conventional td - nrg approach is recovered from eqs .
( [ a - of - t-3 ] ) and ( [ rho - m - of - t ] ) by ( i ) setting @xmath135 ,
( ii ) realizing that @xmath136 for @xmath137 , and ( iii ) adopting the standard nrg approximation @xmath138 , which simplifies @xmath139 to @xmath140 with @xmath141 a natural generalization of the td - nrg is to apply the nrg approximation @xmath142 to the early iterations @xmath143 only , converting eqs . ( [ a - of - t-3 ] ) and ( [ rho - m - of - t ] ) to @xmath144 this equation , which constitutes one of the central results of this paper , interpolates between the td - nrg , corresponding to @xmath135 , and the exact time - dependent expectation value , obtained for @xmath145 .
of course , the latter statement assumes an exact evaluation of @xmath146 , which is an impractical task for @xmath145 .
as discussed below , a proper choice of the parameter @xmath120 allows for an improved evaluation of @xmath146 using alternative methods such as the td - dmrg or cet , with minimal loss of accuracy at the early iterations to which the nrg approximation is applied .
furthermore , by resorting to methods that do not rely on the special structure of the wilson chain to evaluate @xmath146 , one can abandon the exponential decay of the hopping matrix elements beyond site @xmath120 , reducing thereby the discretization errors inherent to the wilson chain .
these principles form the core of the hybrid approach .
we now turn to elaborate on the technicalities of how @xmath120 is selected , the interface with the hybridized method , and the way in which the initial state @xmath147 is constructed . to turn eq .
( [ hybrid ] ) into an operative platform for hybridizing the td - nrg with alternative methods of computing the real - time dynamics of @xmath146 , it is useful to go back to the partitioning of @xmath106 specified in eq .
( [ partitioning - of - psi ] ) and gain a deeper insight into the energy scales encoded into the projectors @xmath110 . applying the operator decomposition of eq .
( [ eqn : operator - projection ] ) to the hamiltonian @xmath104 , the latter is written as @xmath148 where @xmath149 and @xmath150 it is rather easy to see that the different hamiltonian terms that appear in eq .
( [ eqn : h - operator - projection ] ) generally do not commute with one another , i.e. , @xmath151 \neq 0 $ ] and @xmath152\neq 0 $ ] if @xmath153 . according to the nrg philosophy , however , the off - diagonal terms @xmath154 with @xmath155
are expected to be small , as these couple excitations on different energy scales .
consequently , one can approximate @xmath156 with @xmath157 to obtain the approximate hamiltonian @xmath158 evidently , eq . ( [ nrg - approx-1 ] ) becomes exceedingly more accurate the smaller is @xmath120 , acquiring the status of an identity for @xmath145 , since @xmath159 in this case . furthermore , since the hamiltonian terms @xmath160 and @xmath161 with @xmath143 are confined to the subspaces projected out by @xmath162 and @xmath163 , respectively , the hamiltonian of eq .
( [ nrg - approx-1 ] ) is block - diagonal in these subspaces with @xmath164 = [ \hat{h}_m , \hat{h}_{\chi } ] = 0 $ ] .
this allows us to write the time - dependent state @xmath106 within this approximation as @xmath165 where @xmath166 and @xmath167 are the projections of the initial state onto the subspaces defined by @xmath110 and @xmath111 , respectively . in other terms
, it suffices in this approximation to first project out @xmath168 and @xmath169 from the initial state , and then propagate them separately in time , each according to its own hamiltonian .
physically , eq . ( [ decomposition ] ) prescribes a decomposition of the desired time - dependent state into independent components , each associated with a different time scale @xmath170 and evolving according to its own reduced hamiltonian ( either @xmath171 or @xmath172 ) . in the case of spinless electrons the reduced hamiltonian @xmath171 has the explicit form @xmath173 where @xmath174 creates an electron on the @xmath175th site of the wilson chain , @xmath176 is the dimension - full hopping matrix element between sites @xmath175 and @xmath177 along the chain , @xmath71 runs over the nrg eigenstates discarded at iteration @xmath45 , and @xmath178 denotes their corresponding nrg eigenenergies .
note that the projection operators @xmath110 in the right - most term are attached in practice only to @xmath179 and @xmath180 , as all other operators @xmath181 with @xmath182 do not possess any matrix element that takes us out of the subspace defined by @xmath110 .
the hamiltonian @xmath172 is nearly identical , except that the index @xmath45 is replaced with @xmath120 and the discarded states @xmath183 are replaced with the nrg eigenstates retained at the conclusion of iteration @xmath120 . in the presence of additional bands the wilson orbitals @xmath174 acquire an additional flavor index @xmath184 , which may label the spin @xmath185 , an orbital channel @xmath186 , or the spin - channel tuple @xmath187 as in two - channel kondo models ( see , e.g. , ref . ) .
other than setting @xmath188 and adding a suitable summation over @xmath184 , the very same equations carry over to the general multiband case .
the hamiltonian of eq .
( [ h_m - explicit ] ) can be interpreted as modeling a hyper - impurity with the localized configurations @xmath189 and eigenenergies @xmath190 , which are tunnel - coupled to a chain of length @xmath64 .
the size of the impurity is equal to the number of states discarded at iteration @xmath45 .
thus , the calculation of @xmath191 becomes exceedingly more affordable the larger is @xmath45 due to the exponential reduction of the fock space of the chain @xmath56 attached to the hyper - impurity .
we stress , however , that the dimension of the subspace associated with @xmath192 is comparable in size to that of the full wilson chain , hence an accurate evaluation of @xmath193 is similar in complexity to the calculation of the full state @xmath194 .
there is little computational gain in implementing eq .
( [ decomposition ] ) if all components of the wave function must be accurately computed .
fortunately , this generally is not the case for the class of problems of interest , where @xmath147 is some low - lying eigenstate ( typically the ground state ) of an initial hamiltonian @xmath195 . under these circumstances
@xmath147 typically has only negligible overlap with the high - energy states of @xmath104 , i.e. , @xmath196 for the initial nrg iterations .
overlap becomes significant only upon approaching a characteristic energy scale @xmath197 where the spectra of the full and the unperturbed hamiltonians begin to notably deviate from one another .
usually this happens at some characteristic low - energy scale of the problem , e.g. , the kondo temperature @xmath198 in case of the kondo hamiltonian .
consequently , the initial nrg iterations with @xmath143 can be treated using further approximations such as setting @xmath199 in eq .
( [ h_m - explicit ] ) , corresponding to the standard nrg approximation .
since the reduced density matrix @xmath139 requires only the matrix element @xmath200 , one can implement @xmath201 instead of propagating @xmath202 in time , which simplifies the exact result of eq .
( [ a - of - t-3 ] ) to the approximate expression of eq .
( [ hybrid ] ) .
the computational effort can therefore be focused on evaluating @xmath146 , which dominates the expectation value @xmath203 .
most importantly , given the initial state @xmath168 and the effective hamiltonian @xmath204 generated by the nrg , @xmath146 can be computed using one s method of choice . hopping matrix elements decrease according to @xmath205 with @xmath12 , and
all further hopping matrix elements are held constant and equal to @xmath206.,scaledwidth=48.0% ] the physical role of the integer @xmath120 , which so far served as a mere parameter , is now disclosed : it defines the nrg iteration @xmath120 beyond which the time - dependent state should be accurately computed .
moreover , this partitioning can be used to improve the discrete representation of the continuous bath as noted above .
for example , one can design a hybrid chain such that all sites up to @xmath207 are discretized with @xmath12 and all further sites are converted to @xmath19 ( see fig .
[ fig : nrg - tight - binding ] )
. such a chain is impractical for pure nrg - based calculations , but is made possible by resorting to alternative methods for tracking the time evolution of @xmath146 . in this manner discretization errors are significantly reduced at the energy scale @xmath197 , corresponding to the time scale @xmath208 . the crucial point to notice
is that the reduced hamiltonian @xmath209 has the effective bandwidth @xmath210 and acts on a reduced chain of length @xmath211 .
this enables access to long time scales of order @xmath212 using techniques such as the td - dmrg or cet , which otherwise are restricted to far shorter times .
the only remaining uncertainty pertains to a suitable choice of the iteration number @xmath120 . in the absence of a sharp mathematical criterion ,
the choice of @xmath120 should be considered on a case - by - case basis .
qualitatively , one expects the scale @xmath197 to correspond to @xmath213 for the resonant - level model [ see eq .
( [ h - ni - rlm ] ) below ] , and to the kondo temperature @xcite @xmath198 for the kondo model .
the case of an anderson impurity is clearly more subtle , as spin and charge relax on different time scales .
@xcite here a different optimal choice of @xmath120 may apply to observables acting on the spin and charge sectors .
so far , we have assumed the decomposition of the initial state @xmath214 according to eq .
( [ partitioning - of - psi ] ) , but did not specify how @xmath147 is obtained in practice .
the construction of @xmath215 and its projection @xmath168 onto the low - energy subspace defined by @xmath162 depends in detail on the method hybridized with the td - nrg .
our discussion below covers both the td - dmrg and cet .
we begin with the initial nrg run , which provides us with the low - energy hamiltonian @xmath216 corresponding to the initial hamiltonian @xmath195 . here
@xmath217 denotes the projection operator onto the low - energy subspace of @xmath195 retained at the conclusion of iteration @xmath120 . as detailed in eq .
( [ h_m - explicit ] ) , @xmath218 comprises of a hyper - impurity , a residual chain of length @xmath211 , and a tunnel coupling between both parts of the system . in the next step
the ground state @xmath147 of @xmath218 is computed . in case of the td - dmrg
this is done using the standard dmrg algorithm , @xcite while for the cet ( for which a shorter chain @xmath219 is employed ) the davidson method @xcite can be used . at the conclusion of this step one
has the initial state @xmath103 at hand , expressed via the kept nrg eigenstates of @xmath220 : @xmath221 given @xmath147 , the state @xmath168 is obtained by projecting @xmath147 onto the low - energy subspace of the full hamiltonian defined by @xmath162 .
this in turn yields @xmath222 with @xmath223 where @xmath224 is the overlap matrix defined in eq .
( [ eq : s_kr - def ] ) .
this state is then propagated in time according to @xmath225 using either the td - dmrg or cet and fed into eq .
( [ hybrid ] ) . as for the reduced density matrices
@xmath226 entering eq .
( [ hybrid ] ) , these are computed recursively from @xmath147 using the standard td - nrg algorithm .
armed with the hybrid - nrg platform , we proceed in this section to the second major result of this paper the extension of the td - nrg from a single quantum quench to repeated switching events .
coherent control of small nanodevices such as semiconductor quantum dots or superconducting flux qubits can be achieved by applying gate - voltage pulses or time - dependent electromagnetic fields .
for example , circularly polarized laser pulses are used to induce and control the spin polarization in semiconductor quantum dots .
@xcite alternatively , one can apply rapid changes to close - by gate voltages in order to control the energy levels and/or the tunneling rates of a quantum dot .
each of these protocols involves repeated switchings between two distinct configurations of the applied fields , which we denote hereafter by @xmath227 and @xmath228 .
theoretically this corresponds to periodic conversions in time between two quantum - impurity hamiltonians , @xmath20 and @xmath21 , that differ in those components describing the isolated dot and its coupling to the leads .
our goal is to track the time evolution of local expectation values in response to such a sequence of switching events . to clearly formulate the problem
, we assume that the system resides at time @xmath2 in a low - lying eigenstate of the hamiltonian @xmath229 , when its hamiltonian is abruptly converted from @xmath229 to @xmath21 .
the system then evolves in time under the influence of @xmath21 up to time @xmath230 , when the hamiltonian of the system is suddenly switched back to @xmath20 for the duration @xmath231 .
this sequence of switchings is repeated periodically with a period of @xmath232 , as described by the time - dependent hamiltonian @xmath233 ( here @xmath234 ) . within each time interval where the hamiltonian is constant ,
the expectation value of a general local operator @xmath113 is given by eqs .
( [ a - of - t-3 ] ) and ( [ rho - m - of - t ] ) where , depending on the time interval in question , @xmath235 and @xmath236 pertain either to @xmath20 or @xmath21 .
explicitly , @xmath235 is replaced with @xmath237 , where @xmath238 denotes the projection operator of eq .
( [ 1^+_m ] ) written with respect to nrg eigenstates of @xmath239 ( @xmath240 ) . with these adjustments the resulting expression is formally exact , but requires explicit knowledge of the states @xmath241 and @xmath242 . consider a particular time interval @xmath243 in which the system evolves according to the hamiltonian @xmath20 . as discussed in sec .
[ sec : expectation - values ] , the expectation value @xmath203 can be approximated by applying the conventional nrg approximation to the early iterations @xmath143 only .
this yields eq .
( [ hybrid ] ) , where @xmath235 , the eigenenergies @xmath244 and @xmath245 , and the state labels @xmath44 and @xmath43 correspond to @xmath20 .
the only formal modification as compared to eq .
( [ hybrid ] ) pertains to the reduced density matrix @xmath246 , in which the initial state @xmath215 must be replaced with @xmath247 at time @xmath248 . the same set of rules carry over to any given time interval @xmath249 , except that @xmath20 is replaced with @xmath21 , and the initial state @xmath215 is replaced with @xmath247 at time @xmath250 .
this leaves us with those instances in time where the hamiltonian is abruptly converted from @xmath20 to @xmath21 or vice versa . for concreteness
let us focus on the time instance @xmath250 , when the hamiltonian is converted from @xmath20 to @xmath21 .
as described in eq .
( [ partitioning - of - psi ] ) , the state of the system can be formally decomposed within the time interval @xmath243 into @xmath251 with @xmath252 . here
the states @xmath253 are projected according to the nrg eigenstates of @xmath20 .
the time - dependent density operator is next divided into @xmath254 , where @xmath255 and @xmath256 . setting @xmath257 and replacing @xmath258 in eq .
( [ eqn : reduced - dm - def ] ) , the reduced density matrix with respect to the eigenstates of @xmath229 reads @xmath259 where we have defined @xmath260 an equivalent expression @xmath261 applies to the reduced density matrix with respect to the nrg eigenstates of the hamiltonian @xmath21 . in the notations of ref .
[ see eqs.(31 ) and ( 32 ) therein ] , the first term @xmath262 in eq .
( [ rho - red - via - chi ] ) is of the pure form @xmath263 , having none of the components @xmath264 , @xmath265 , and @xmath266 .
thus , one has the exact conversion @xcite @xmath267 where @xmath268 is the overlap matrix between the nrg eigenstates of @xmath20 and @xmath21 defined by eq .
( [ eq : s_kr - def ] ) .
by contrast , the second term @xmath269 in eq .
( [ rho - red - via - chi ] ) involves all four components @xmath270 , @xmath271 , @xmath272 , and @xmath273 , and as a result lacks an explicit relation @xcite to @xmath274 .
similar to eq .
( [ trans - of - chi ] ) , we approximate @xmath274 by @xmath275 leading to the compact transformation rule @xmath276 the complementary transformation rule for switching from @xmath21 to @xmath20 reads @xmath277 it should be emphasized that eq .
( [ delta - rho - transformed ] ) is an _ ad - hoc _ approximation , whose quality is difficult to assess _ a - priori_. its accuracy is necessarily controlled by the smallness of @xmath278 , which in turn is reduced the smaller is @xmath120 .
below we present numerical results demonstrating this point .
in addition to the reduced density matrix , eq .
( [ hybrid ] ) requires explicit knowledge of the projected state , either @xmath279 or @xmath280 , depending on the time interval in question .
it is therefore necessary to formulate the transformation rule of @xmath281 upon conversion of the hamiltonian .
focusing again on the time instance @xmath282 , the state @xmath283 is formally given by @xmath284 where @xmath285 and @xmath286 are projected according to the operators @xmath287 and @xmath288 pertaining to the hamiltonian @xmath20 .
the right - most term in eq .
( [ chi_2n - switch ] ) has the exact representation @xmath289 which follows from the fact that @xmath290 vanishes by construction for all @xmath291 . as for the remaining terms on the right - hand side of eq .
( [ chi_2n - switch ] ) , these require explicit knowledge of the states @xmath285 .
unfortunately , it is unfeasible to keep track of the states @xmath292 , which forces yet another approximation . our strategy is to omit these terms altogether , with the expectation that their combined contribution is typically small .
naturally , the quality of this approximation will depend on the choice of @xmath120 and on details of @xmath20 and @xmath21 .
this leaves us with the approximate transformation rule @xmath293 along with its counterpart @xmath294 we are now in position to summarize the proposed algorithm for computing @xmath295 for a sequence of switching events .
we begin by rewriting eq .
( [ hybrid ] ) in the form @xmath296 where @xmath186 equals @xmath297 or @xmath298 depending on the time interval , and @xmath299 is the matrix representation of @xmath113 at iteration @xmath45 with respect to the nrg eigenstates of @xmath239 .
the first step is to evaluate @xmath300 and @xmath301 using the td - nrg methodology detailed in ref . . from this point on these quantities
are evolved in time according to the following set of rules : + ( i ) at time @xmath302 we switch from @xmath303 to @xmath304 by setting @xmath305 ( ii ) in the time interval @xmath306 , the density matrices and wave function are propagated in time according to @xmath307 where @xmath308 is the projected low - energy hamiltonian corresponding to @xmath21 . + ( iii ) at time @xmath309 we switch back from @xmath304 to @xmath303 by setting @xmath310 ( iv ) the density matrices and wave function are propagated in the time interval @xmath311 according to @xmath312 where @xmath313 is the projected low - energy hamiltonian corresponding to @xmath20 .
we stress that four distinct approximations enter eqs .
( [ chi_m - switching-1])([eqn : chi - rho - evolution-2 ] ) : ( i ) a discrete representation of the continuous bath ; ( ii ) the standard nrg approximation @xmath314 that is used to propagate the reduced density matrices within each time interval where @xmath104 is fixed ; ( iii ) the transformation rule of eq .
( [ delta - rho - transformed ] ) for @xmath315 at each switching event ; and ( iv ) the omission of those terms that originate from the @xmath316 s in eq .
( [ chi_2n - switch ] ) in the transformation rule for @xmath235 at each switching event .
while the first two approximations are general in nature and apply , in particular , to the td - nrg , the latter two approximations are specific to the present formulation of repeated switchings .
the quality of those approximations , which become exact for @xmath145 , can be tested _ a - posteriori _ by comparison to cases where exact solutions are available , as done in sec .
[ sec : rlm - switching ] for the noninteracting resonant - level model .
our presentation thus far was quite general and did not specify a particular method to be hybridized with the td - nrg . as a proof of principle , we shall demonstrate in sec .
[ sec : periodic - switching ] the hybridization of the td - nrg with the cet , whose principles and implementation are reviewed below . the cet @xcite offers an accurate way to calculate the time evolution of an initial state @xmath103 under the influence of a general stationary finite - dimensional hamiltonian @xmath104 : @xmath317 the main idea of the method is to construct a stable numerical approximation for the time - evolution operator @xmath318 that is independent of the initial state @xmath215 and whose error can be reduced to machine precision .
its limitation lies in the need to explicitly store certain states in the course of the calculation , which limits the size of the hilbert space that can be handled .
there are different ways to expand the time - evolution operator , the most direct one being the conventional expansion of the exponent in powers of @xmath104 .
one would like , however , to use an expansion that converges uniformly , independent of the initial state @xmath215 .
a suitable choice are the chebyshev polynomials , defined by the recursion relation @xmath319 subject to the initial conditions @xmath320 and @xmath321 .
as is well known , the chebyshev polynomials can be used to expand any function @xmath322 on the interval @xmath323 .
explicitly , @xmath322 is expressed as an infinite series @xmath324 where the expansion coefficients @xmath325 are given by @xmath326 the expansion in terms of chebyshev polynomials can be equally applied to any function @xmath327 with an arbitrary support @xmath328 using the transformation @xmath329 which maps the interval @xmath330 onto @xmath331 . in doing
so one expands in practice the function @xmath332 . using the rules laid above
, the function @xmath333 is expanded for @xmath334 as @xmath335 with @xmath336 here @xmath337 are the bessel functions , @xmath338 equals @xmath339 , and @xmath340 .
accordingly , the time - evolution operator @xmath341 is expanded as @xmath342 where [ b_n ] @xmath343 here @xmath344 ( @xmath345 ) is the maximal ( minimal ) eigenenergy of @xmath104 , @xmath346 equals @xmath347 , @xmath348 , and @xmath349 is the `` transformed '' hamiltonian @xmath350 finally , applying eq .
( [ chebyshev - exp - e^-ih ] ) to the initial state @xmath215 one obtains @xmath351 where the infinite set of states @xmath352 obey the recursion relation@xcite @xmath353 subject to the initial condition @xmath354 and @xmath355 .
several comments are in order .
first , all time dependence is confined in eq .
( [ cet ] ) to the expansion coefficients @xmath356 of eqs .
( [ b_n ] ) , which are independent of the initial state @xmath215 .
second , the chebyshev recursion relation of eq .
( [ chebyshev - recursion - relation ] ) reveals the iterative nature of the calculations . starting form the initial state @xmath215
, one constructs all subsequent states @xmath357 using repeated applications of the `` transformed '' hamiltonian @xmath349 . note that in practice only two such states need be stored in memory at each given time .
third , since @xmath358 for large order @xmath175 , the chebyshev expansion converges quickly as @xmath175 exceeds @xmath359 . finally , the chebyshev expansion has the virtue that numerical errors are practically independent of @xmath360 , allowing access to extremely long times .
the main limitation of the approach , as commented above , stems from the size of the hilbert space , since each of the states @xmath357 must be constructed explicitly . in our applications of the approach ( where @xmath168 serves as the initial state ) , this hilbert space comprises of the kept nrg states at iteration @xmath120 typically of the order of @xmath361 and the remaining chain @xmath219 , whose dimension is @xmath362 ( @xmath363 being the number of distinct configurations of a single spinless site ) . as a result ,
application of the cet is confined to rather short chains that can not be used to access arbitrarily long time scales .
as already pointed out in the introduction , the wilsonian discretization of the continuous bath significantly influences the quench dynamics , independent of which finite - size approach exact diagonalization , td - nrg , or the td - dmrg is used to track the real - time dynamics of the system . as illustrated in fig .
[ fig : rlm - n - vs - t ] , deviations from the exact continuum - limit result stem from current reflections at sites along the wilson chain , caused by the exponentially decreasing hopping matrix elements that are used .
these in turn produce an exponentially decreasing transport velocity . in this section
, we present a preliminary discussion aimed at demonstrating the potential of substituting the standard wilson chain with a hybrid chain of the type depicted in fig .
[ fig : nrg - tight - binding ] .
we shall do so by investigating a single quantum quench in the noninteracting resonant - level model ( rlm ) , which can be solved exactly on essentially any finite - size chain using exact diagonalization of the single - particle eigenmodes .
the availability of an exact analytical solution for the real - time dynamics of the level occupancy in the continuum limit [ see eq .
( 50 ) of ref . ]
makes this model an ideal benchmark for testing the quality of different hybrid chains in reproducing the continuum - limit result .
thus , one can clearly separate deviations caused by the structure of the chain from those that originate from further approximations underlying the td - dmrg or td - nrg .
furthermore , the rlm sets the stage for more complicated interacting models , such as the interacting resonant - level model discussed below .
physically , the rlm describes the coupling of a single spinless level to a continuous band of width @xmath364 via the single - particle hopping matrix element @xmath365 : @xmath366 here @xmath367 creates an electron on the level , @xmath368 creates a band electron with momentum @xmath72 , and @xmath369 labels the number of distinct values of @xmath72 .
the relevant energy scales in the problem include the level energy @xmath370 , along with the hybridization width @xmath371 , where @xmath372 is the conduction - electron density of states at the fermi level . in the rlm ( solid black line ) and its exact numerical evaluation for various hybrid chains of the type depicted in fig .
[ fig : nrg - tight - binding ] .
the full red line depicts the case of a pure wilson chain of length @xmath8 .
panel ( a ) : keeping the total chain length fixed at @xmath8 and varying the partitioning parameter @xmath120 .
panel ( b ) : for @xmath373 and two different chain lengths @xmath8 and @xmath374 .
model parameters : @xmath375 , @xmath376 , @xmath6 , and @xmath377 .
, scaledwidth=48.0% ] in the rlm ( solid black line ) and its exact numerical evaluation for various hybrid chains of the type depicted in fig .
[ fig : nrg - tight - binding ] .
the full red line depicts the case of a pure wilson chain of length @xmath8 .
panel ( a ) : keeping the total chain length fixed at @xmath8 and varying the partitioning parameter @xmath120 . panel ( b ) : for @xmath373 and two different chain lengths @xmath8 and @xmath374 .
model parameters : @xmath375 , @xmath376 , @xmath6 , and @xmath377 .
, scaledwidth=48.0% ] we are interested in tracking the real - time dynamics of the level occupancy @xmath378 in response to a sudden quench from @xmath379 to @xmath380 .
figure [ fig : rlm - hybrid - chain - quench ] depicts the time evolution of @xmath9 after the level energy has been abruptly shifted from @xmath381 to @xmath382 while keeping the hybridization width fixed at @xmath383 .
we used the same model parameters as in fig .
[ fig : rlm - n - vs - t ] , but varied the chain structure by considering different values of the partitioning parameter @xmath120 and different chain lengths @xmath11 .
explicitly , as illustrated in fig .
[ fig : nrg - tight - binding ] , these chains comprise of an initial wilson chain with exponentially decreasing hopping matrix elements @xmath384 up to @xmath207 , converting to a constant hopping matrix element @xmath385 for @xmath386 . as can be seen in fig .
[ fig : rlm - hybrid - chain - quench](a ) , there is a systematic improvement in the agreement with the continuum - limit result upon decreasing @xmath120 . specifically ,
for @xmath373 the deviations remain quite small up to @xmath387 , at which point there is a revival of @xmath9 which nearly reaches its original value @xmath388 for @xmath389 .
in contrast to a pure wilson chain , the current velocity becomes a constant along the chain sites with @xmath390 , hence current reflections occur only at the end of the chain .
@xcite the time at which the revival of @xmath9 is observed is simply given by the round - trip time for a charge pulse that originates from the impurity to reach the chain end and return .
note that this time is shorter for @xmath373 than for @xmath391 .
since charge is a globally conserved quantity , true thermalization and relaxation can occur only in the thermodynamic limit @xmath17 .
the deviation of the total charge from its thermalized value following the quench is simply given by the difference in the equilibrated charges before and after the quench . since the change in total charge is @xmath392 for such a local quench , there is an @xmath393 contribution to each reservoir degree of freedom which can be safely neglected in the thermodynamic limit . for any finite - size chain , however , this effect remains finite and relevant for the long - time limit .
in particular , the continuity equation leads to charge reflections at the end of the tight - binding chain @xcite such that the round - trip time is controlled by the chain length .
indeed , in fig .
[ fig : rlm - hybrid - chain - quench](b ) we compare @xmath9 for two hybrid chains , each partitioned at @xmath373 .
the two chains share the same characteristic energy scale @xmath197 for their partitioning , but one has a total length of @xmath8 and the other @xmath394 .
while the short - time dynamics is nearly identical and agrees well with the continuum - limit result , the revival time for the long hybrid chain is pushed way beyond the time interval presented in fig .
[ fig : rlm - hybrid - chain - quench ] .
thus , discretization errors have been nearly eliminated on the time scale of interest .
this latter example vividly illustrates the potential of hybridizing the td - nrg with the td - dmrg , as the tight - binding chain length involved is of typical dmrg size . since in our example
the effective bandwidth @xmath197 is of order @xmath383 , the td - nrg generates an effective hamiltonian @xmath395 whose effective bandwidth @xmath197 can be of several orders of magnitude smaller than the bare conduction - electron bandwidth @xmath396 .
this in turn allows access to long time scales of order @xmath397 which lie beyond the reach of pure td - dmrg , while significantly reducing discretization errors that are inherent to the td - nrg .
a concluding word is in order about the optimal choices of @xmath120 and the structure of the hybrid chain .
the answers to these questions are quite difficult and can be expected to depend both on the model and observable of interest .
for instance , spin and charge excitations generally propagate with different velocities .
it is therefore quite feasible that the optimal choices of @xmath120 and @xmath11 will dependent on the observable in question .
a more systematic study of the optimal hybrid chain is left for future research .
so far , we have stressed the potential of using hybrid chains by comparing their exact numerical solutions with the continuum - limit result for simple quenches . in this section ,
we extend our focus to the hybrid - nrg approach as formulated in sec .
[ sec : switchings ] for periodic switching .
to benchmark our switching algorithm , we compare it with exact diagonalization solutions on a pure wilson chain , postponing further discussion of the usage of hybrid chains and the comparison to exact continuum - limit results .
therefore , all calculations are performed on a standard wilson chain with identical parameters , hybridizing the cet with the td - nrg at a given site @xmath120 . in order to illustrate how the effective low - energy hamiltonian generated by the nrg can be fed into the cet ( or into any other method of choice for that matter ) , we consider an extreme wide - band limit by setting the bare bandwidth of the rlm to @xmath398 . using the notations introduced in sec .
[ sec : switchings ] , we start at time @xmath2 with a system that resides in the ground state of @xmath20 , and switch repeatedly between @xmath20 and @xmath21 after each additional time interval @xmath399 . as our basic energy scale
we set @xmath400 , whose associated time scale @xmath401 lies many orders of magnitude beyond the reach of the cet when applied directly to the full wilson chain .
we work with a chain of fixed length @xmath402 and the logarithmic discretization parameter @xmath403 , such that @xmath404 . to initiate the calculation
, we first perform an nrg run for the initial hamiltonian @xmath20 and select the ground state of the @xmath11th iteration as our initial state . in this case
one can use the davidson method @xcite to construct the initial state , as discussed in sec .
[ sec : initial - psi-0 ] . ] @xmath103 ( for even @xmath11 the ground state is unique ) .
we then perform a second nrg run for the other hamiltonian @xmath21 , during which we construct the overlap matrices @xmath268 .
technically these two nrg runs can be performed in parallel , in which case the overlap matrices @xmath268 are calculated at the end of each nrg iteration .
backtracking from iteration @xmath11 to iteration @xmath65 , the reduced density matrices @xmath405 are computed using the standard td - nrg algorithm.@xcite these density matrices , along with the projected state @xmath406 , are then used as a seed for the hybrid - nrg algorithm detailed in eqs .
( [ a(t)-switching])([eqn : chi - rho - evolution-2 ] ) . before discussing the full hybrid approach ,
let us focus for a moment on a simple extension of the td - nrg to periodic switching , corresponding to setting @xmath135 in eqs .
( [ a(t)-switching])([eqn : chi - rho - evolution-2 ] ) .
this implies taking @xmath407 throughout the calculation . using @xmath408 and @xmath409 , fig .
[ fig : switching5 ] depicts a comparison of the td - nrg with an exact diagonalization solution on a wilson chain with @xmath410 .
since the decay rate is given by @xmath383 , the system almost fully equilibrates before the next switching event takes place . as a result the deviations of the periodic td - nrg from the exact numerical solution are surprisingly small .
in particular , the periodic td - nrg correctly tracks the general structure of the precise solution .
nevertheless , a systematic degradation in accuracy is observed upon going from one switching cycle to the next .
this loss of accuracy can be quantified by the growing discontinuity @xmath411 at successive switching events , caused by the approximation made to @xmath412 .
reduction of the switching time to @xmath413 prevents the system from relaxing to a new equilibrium state . in this case
high - energy excitations , which are cut off by neglecting the three additional terms @xmath271 , @xmath272 , and @xmath273 , contribute significantly to the time evolution of @xmath414 . since switching occurs on a shorter time scale energy is frequently pumped into the system , exciting it in such a way that the three neglected high - energy contributions to @xmath412 gain increasing importance with time .
indeed , the periodic td - nrg result for @xmath9 , depicted by the blue curve in fig .
[ fig : switching1](b ) , shows a sizable accumulated error that grows systematically in time .
although the simple periodic extension of the td - nrg already captures the correct short - time dynamics , the externally driven nonequilibrium state increasingly deviates with time from the transient dynamics of a single quantum quench .
energy is locally added and removed from the system , which can be partially dissipated via heat current flowing between the impurity and the finite - size bath . to properly capture this physics
, we next employ the hybrid approach to periodic switching .
the key idea is to identify a suitable low - energy subspace that is large enough for the neglected contributions in our approximation to be small . a trivial limit is given by setting @xmath415 , for which the hybrid approach reproduces by construction the exact solution on the finite - size chain .
even though this limit has no practical value , it illustrates the point that a reduction in @xmath120 should systematically improve the quality of the results . with this understanding ,
we extend our discussion of the hybrid approach to @xmath416 . at each time interval
where @xmath104 is fixed we evolve @xmath242 in time using the cet , which is essentially exact on all time scales of interest .
the cet is restricted , however , in the size of the hilbert space one can treat , which limits the length of the residual chain @xmath219 extending beyond site @xmath120 . with our machines we can comfortably access up to @xmath417 basis states , above which the computational effort rapidly becomes too exhaustive .
keeping @xmath418 states at the conclusion of each nrg iteration , this sets an upper limit of @xmath419 addition wilson sites beyond site @xmath120 , i.e. , @xmath420 . to reduce finite - size effects we average over @xmath421 equally distributed values of the twist parameter @xcite @xmath422 $ ] .
below we present results for different values of @xmath423 and @xmath424 .
but for @xmath425 and @xmath426 ( i.e. , no @xmath427 averaging ) .
note the much stronger @xmath120 dependence as compared to @xmath408.,title="fig:",scaledwidth=48.0% ] but for @xmath425 and @xmath426 ( i.e. , no @xmath427 averaging ) .
note the much stronger @xmath120 dependence as compared to @xmath408.,title="fig:",scaledwidth=48.0% ] figures [ fig : switching5 ] and [ fig : switching1 ] show the time evolution of @xmath9 in response to repeated switchings between @xmath428 and @xmath429 , keeping @xmath430 fixed .
figure [ fig : switching5 ] depicts four successive switch events with the time interval @xmath408 , while fig .
[ fig : switching1 ] shows ten successive switch events with the shorter time interval @xmath425 . for comparison ,
the exact time evolutions on the wilson chain are depicted by the black lines , after averaging over the different values of @xmath427 .
these solutions are obtained by exact diagonalization of the single - particle eigenmodes of each hamiltonian and using them to propagate @xmath431 in time .
several points are noteworthy . upon decreasing @xmath120
, the hybrid nrg - cet curves gradually converge onto the exact result for the wilson chain , both for @xmath425 and @xmath408 .
specifically , by @xmath432 the hybrid nrg - cet approach essentially reproduces the exact curves .
the largest deviations are usually found immediately following a switching event , when a discontinuity generally develops in the occupancy produced by the hybrid nrg - cet approach .
as noted above , the discontinuity stems from the approximations employed in deriving eqs .
( [ chi_m - switching-1])([eqn : chi - rho - evolution-2 ] ) .
it decreases in size with decreasing @xmath120 , reflecting a reduction in the accumulated weight of the @xmath433 s in favor of @xmath235 in the expansion of eq .
( [ partitioning - of - psi ] ) .
indeed , the smaller the accumulated weight of the @xmath433 s the more accurate are the approximations employed in deriving the hybrid - nrg . for @xmath434
, the hybrid approach corresponds to the simple extension of the td - nrg to periodic switching . as can be seen in fig .
[ fig : switching1](b ) , the discontinuities at @xmath435 are particularly large for the shorter switching time @xmath425 since more energy is pumped into the system .
these discontinuities are greatly reduced for the longer time interval @xmath408 as depicted in fig .
[ fig : switching5 ] . as discussed above ,
this behavior is correlated with the fact that the occupancy @xmath9 for @xmath408 has nearly decayed in each time interval to its new equilibrium value corresponding to the hamiltonian @xmath239 in that time segment . in other terms
, the state of the system behaves as if it has effectively decayed to the new ground state , leaving behind only weak reminiscence of the transient behavior . and
@xmath436 for fixed @xmath437 .
all other parameters are the same as in fig .
[ fig : switching5 ] . , scaledwidth=48.0% ] figure [ fig : switching - of - gamma ] shows the complementary response to repeated switchings between two different hybridization widths , @xmath438 and @xmath436 for fixed @xmath437 . as can be seen , the discontinuities at @xmath435 are barely visible in this case , yet discretization errors do give rise to high - frequency wiggles that are essentially absent in figs .
[ fig : switching5 ] and [ fig : switching1 ] .
the usage of @xmath427-averaging is essential for reducing these discretization effects , both at the level of the hybrid - nrg and in the framework of exact diagonalization .
similar to the case of a single quantum quench , discretization errors become more and more pronounced the larger are the deviations between @xmath20 and @xmath21 and the longer the time that has elapsed . clearly , the nature of the perturbation ( e.g. , @xmath439 vs @xmath440 ) is also of importance .
it should be emphasized , however , that @xmath427-averaging alone is insufficient for removing low - frequency oscillations that appear when the deviations between @xmath20 and @xmath21 are large . to summarize this subsection ,
we have demonstrated how the nrg can be used to systematically construct effective low - energy hamiltonians @xmath313 and @xmath308 whose bandwidth is smaller by orders of magnitude than the bare conduction - electron bandwidth @xmath396 of the original model .
as a result methods such as the td - dmrg , whose accuracy is usually confined to times of order @xmath441 , can be used within the hybrid platform to access exponentially long time scales .
this follows from the fact that the new effective bandwidth @xmath442 can be exponentially smaller than @xmath396 .
having established the accuracy of the hybrid nrg - cet approach for the noninteracting rlm , we proceed to apply it to an interacting problem that lacks an exact reference solution .
specifically , we shall use the hybrid nrg - cet to track the real - time dynamics of the interacting resonant - level model ( irlm ) in a regime not accessible to other available methods . in the irlm , @xcite the resonant - level model of eq .
( [ h - ni - rlm ] ) is supplemented by a local contact interaction @xmath26 between the level and the band electrons : @xmath443 here @xmath444 stands for normal ordering with respect to the filled fermi sea .
physically , the contact interaction @xmath26 accounts for the local capacitive coupling between the localized level @xmath367 and the band electrons .
the total hamiltonian of the model is given by @xmath445 where @xmath446 represents the rlm hamiltonian of eq .
( [ h - ni - rlm ] ) . at low energies , the irlm is equivalent to a renormalized noninteracting rlm , both describing a phase - shifted fermi liquid.@xcite at resonance the model is characterized by the renormalized tunneling rate @xmath447 where @xmath371 is the noninteracting hybridization width of the rlm defined in eq .
( [ h - ni - rlm ] ) , @xmath186 equals @xmath448 , and @xmath449 is the scattering phase shift associated with @xmath26 alone ( namely , in the absence of @xmath365 ) . within the nrg ,
@xmath450 is conveniently extracted from the zero - temperature charge susceptibility @xmath451 , evaluated at @xmath13 .
explicitly , we adopt the working definition @xmath452 . in fig .
[ fig : switching - with - u ] we show the time evolution of @xmath9 in response to repeated switchings between @xmath453 and @xmath454 for @xmath455 .
the bare hybridization strength @xmath456 was adjusted numerically so as to maintain a fixed @xmath457 .
the effect of a finite @xmath26 is two - fold .
first , it renormalizes the bare hybridization width from @xmath383 to @xmath450 , which sets the basic time scale in the problem : @xmath458 .
second , there are pronounced oscillations that are absent for @xmath459 , having the characteristic period @xmath460 .
to illustrate this point , we have borrowed from fig .
[ fig : switching5 ] the exact time evolution of @xmath9 on the wilson chain for @xmath459 and @xmath461 .
the effect of @xmath26 clearly goes beyond just a simple renormalization of the parameters of the noninteracting rlm , generating new interaction - induced oscillations . to further investigate this point ,
we have plotted in fig .
[ fig : quench - with - u ] the real - time dynamics in response to a single quantum quench from @xmath462 to @xmath454 , for different values of the coulomb repulsion @xmath26 .
the bare hybridization width @xmath430 was adjusted separately for each value of @xmath26 so as to maintain a fixed @xmath463 .
all curves begin at essentially the same initial occupancy @xmath464 and decay at long time to @xmath465 .
hence the low - energy fixed point , which governs the thermodynamics , is fully determined by @xmath450 and @xmath370 .
the intermediate dynamics , on the other hand , is quite sensitive to @xmath26 .
starting from @xmath459 and increasing @xmath26 , damped oscillations gradually develop with a characteristic period that only weakly depends on @xmath26 .
in contrast , the amplitude of the oscillations and their damping rate strongly depend on @xmath26 .
the larger is @xmath26 the slower does the envelope function decays to zero , resulting in the emergence of two distinct time scales for large @xmath26 : the period of oscillations @xmath466 and the characteristic damping time @xmath467 .
for example , while @xmath468 for @xmath469 , the corresponding damping time is of order @xmath470 . to understand the origin of these interaction - induced oscillations , it is instructive to consider the limit of a large coulomb repulsion , @xmath471 . in this extreme the level degree of freedom @xmath367 and the zeroth wilson shell @xmath472 decouple from the rest of the chain , being confined to a total valence of one : @xmath473 [ this valence is fixed by the normal - ordered form of @xmath474 as defined in eq .
( [ eqn : irlm - hu ] ) ] .
the corresponding subspace comprises of the two configurations @xmath475 and @xmath476 , where @xmath477 denotes the state in which the resonant level and the zeroth wilson shell are both empty . omitting the coupling to the rest of the chain
, the problem has been reduced in effect to a @xmath478 matrix whose eigenenergies are @xmath479 consequently , @xmath9 must display quantum beats with the frequency @xmath480 to make contact with fig .
[ fig : quench - with - u ] , it is necessary to express the period @xmath481 in terms of @xmath482 .
a straightforward calculation gives @xmath483 resulting in @xmath484 .
substituting @xmath485 into eq .
( [ omega - for - large - u ] ) then yields @xmath486 finally , setting @xmath487 to match the value used in the curves of fig .
[ fig : quench - with - u ] one obtains @xmath488 , in excellent agreement with the period observed in fig .
[ fig : quench - with - u ] .
our strong - coupling analysis clearly reveals that @xmath26 enters the quantum - beat frequency only via @xmath450 .
however , @xmath26 strongly influences the amplitude of the oscillations and their damping rate .
in contrast to the period of oscillations which is well described by @xmath471 , the damping time @xmath467 requires a finite coupling to the rest of the chain . for @xmath489
the impurity and the first wilson shell decouple from the rest of the chain , resulting in coherent quantum beats between the two singly occupied eigenstates .
once @xmath26 becomes finite the system can decay incoherently to the lowest of the two singly occupied states through a residual coupling to the rest of the chain .
it is this decay that determines @xmath467 . since the residual coupling should scale as @xmath490 for large interactions , @xmath467 should scale as @xmath491 for @xmath492 .
this corresponds to a damping rate that falls off as @xmath493 .
in this paper , we have extended the td - nrg in two ways .
first , we devised a platform for combining the td - nrg with complementary methods such as the td - dmrg and cet for tracking the real - time dynamics of quantum - impurity systems .
second , we extended the td - nrg from its original realm of a single quantum quench to more complicated forms of driven dynamics where repeated switchings are applied to the system . as a proof of principle we combined the td - nrg with the cet to compute the response of a resonant level , both with and without interactions , to repeated switchings of its energy level and tunneling amplitude to the band .
such a model can be used to describe a single coulomb - blockade resonance in small quantum dots in regimes where spin degeneracy is unimportant . in the absence of interactions
we have critically examined our new approach by detailed comparisons to an exact evaluation of the time evolution on the wilson chain . as long as the perturbations are not too large , good accuracy
is maintained over a fairly large number of switching events .
usage of the hybrid nrg - cet greatly improves the accuracy in cases where large deviations develop between the exact curve and the one produced without invoking the cet , see , e.g. , fig .
[ fig : switching1 ] .
our present implementation of the hybrid approach is subject to the inherent restriction of the cet to rather small finite - size systems .
we expect a significant boost in accuracy and flexibility by combining the td - nrg with the td - dmrg , which is capable of treating far larger systems .
in particular , our approach opens the possibility to boost the td - dmrg to exponentially long time scales by using the td - nrg to ( i ) construct an effective low - energy hamiltonian and ( ii ) account for the short - time dynamics .
this portion of the research is left for future work .
the potential of the new hybrid approach was next demonstrated by applying it to repeated switchings in the irlm .
apart from specially tuned models , @xcite this constitutes to our knowledge the first study of such driven dynamics in interacting quantum impurity systems .
although equivalent at sufficiently low energies to a noninteracting rlm , the driven dynamics of the irlm shows clear traces of the interaction . in particular , interaction - induced oscillations develop for strong coulomb repulsion , which have no equivalent in the absence of interactions .
we were able to explain the period of oscillations and its weak dependence on @xmath26 by invoking the limit @xmath471 .
the associated damping time scales as @xmath491 , and can greatly exceed the natural damping time @xmath494 of the noninteracting rlm .
such a long damping time is strictly the effect of interactions .
we are grateful to dotan goberman , achim rosch , sebastian schmitt and matthias vojta for helpful discussions .
this work was supported by the german - israeli foundation through grant no .
1035 - 36.14 , by the israel science foundation through grant no .
1524/07 ( e.e . and a.s . ) , and by the deutsche forschungsgemeinschaft under an 275/6 - 2 ( f.g .
and f.b.a ) .
60ifxundefined [ 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 _ _ ( , ) link:\doibase 10.1103/physreva.57.120 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.64.849 [ * * , ( ) ] @noop _ _ , , vol .
( , , ) @noop _ _ ( , , ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.81.165106 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1140/epjst / e2009 - 00962 - 3 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/revmodphys.78.275 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.101.066804 [ * * , ( ) ] http://arxiv.org/pdf/0803.3004 [ * * , ( ) ] http://stacks.iop.org/0953-8984/13/i=44/a=314 [ * * , ( ) ] link:\doibase 10.1103/physrevb.73.045117 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.49.13929 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/17/35/008 [ * * , ( ) ] http://stacks.iop.org/1742-6596/220/i=1/a=012022 [ * * , ( ) ] link:\doibase 10.1103/physrevb.22.613 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.78.201301 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.81.153303 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.25.4815 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.144301 [ * * , ( ) ] | the time - dependent numerical renormalization - group approach ( td - nrg ) , originally devised for tracking the real - time dynamics of quantum - impurity systems following a single quantum quench , is extended to multiple switching events .
this generalization of the td - nrg encompasses the possibility of periodic switching , allowing for coherent control of strongly correlated systems by an external time - dependent field . to this end , we have embedded the td - nrg in a hybrid framework that combines the outstanding capabilities of the numerical renormalization group to systematically construct the effective low - energy hamiltonian of the system with the prowess of complementary approaches for calculating the real - time dynamics derived from this hamiltonian .
we demonstrate the power of our approach by hybridizing the td - nrg with the chebyshev expansion technique in order to investigate periodic switching in the interacting resonant - level model .
although the interacting model shares the same low - energy fixed point as its noninteracting counterpart , we surprisingly find the gradual emergence of damped oscillations as the interaction strength is increased .
focusing on a single quantum quench and using a strong - coupling analysis , we reveal the origin of these interaction - induced oscillations and provide an analytical estimate for their frequency .
the latter agrees well with the numerical results . |
vibrationally resolved electronic spectra ( _ e.g. _ one - photon absorption and emission spectra ) are within the born - oppenheimer framework usually interpreted in terms of franck - condon ( fc ) factors ( fcfs ) @xcite .
accordingly , one can try to obtain the shape of the spectral profile for a fc - allowed transition from computed fcfs in frequency domain .
however , the evaluation of fcfs for large molecular systems is challenging even within the harmonic approximation if one has to take duschinsky mode mixing ( rotation ) @xcite into account .
this is because multi - variate hermite polynomials have then to be evaluated for each fc integral , rather than only uni - variate hermite polynomials as is the case for the comparatively simple parallel harmonic oscillator model .
the computational task becomes more difficult as the molecular size and temperature increases because the number of fc integrals grows vastly . to describe fc - forbidden or weakly allowed transitions ,
one has to go beyond the condon approximation and employ for instance a herzberg - teller ( ht ) expansion @xcite of the electronic transition moment with respect to the normal coordinates . as a result ,
the calculation of the vibronic spectrum for a non - condon process is even more difficult than for a fc - allowed transition because one has to evaluate many matrix elements of the non - condon operators which require for each ht integral in general the calculation of combinations of several fc integrals .
the number of fc integrals and matrix elements of non - condon operators to be evaluated in a sum - over - states approach can be significantly reduced with the help of rigorous integral prescreening strategies @xcite . however , this time - independent ( ti ) calculation of the spectral profile in frequency domain is still considerably more expensive than an alternative time - dependent ( td ) approach that exploits time - correlation functions ( tcfs ) ( see _ e.g. _ ref .
@xcite ) , but offers the ability to directly assign individual peaks in the spectrum . as we have outlined earlier @xcite , a unified coherent state - based generating function ( csgf ) approach @xcite can be used both for rigorous integral prescreening strategies and tcf calculations which combine the strengths of both approaches and complement each other favourably . even in the less demanding td approach , however , one usually has to invest significant computational time for often unnecessary spectral details .
cumulants ( or moments ) of a distribution ( see _ e.g. _ refs .
@xcite ) can deliver highly useful information . from this one can either attempt to reconstruct the spectral shape or try to estimate the relevant spectral profile , which can be exploited in subsequent ti and td approaches @xcite .
cumulants of the vibronic spectrum can be obtained from the csgf directly without computing the total spectrum in frequency domain .
this method was exploited already in ref .
@xcite for fc - allowed transitions , and we report herein an extension of this method to incorporate non - condon transitions . as application , we present the profile of the @xmath1 transition of benzene , which is in the electric dipole approximation franck - condon forbidden , at various temperatures within the linear ht approximation and compare the cumulant expansion result with the tcf approach .
the spectral profile ( @xmath2 ) can be expressed via the fourier transform ( ft ) of the tcf ( @xmath3 ) that depends on the time @xmath4 and temperature @xmath5 , namely @xmath6 where @xmath7 is the transition energy and @xmath8 the @xmath9 transition energy .
the corresponding occupancy representation for the tcf can be obtained from fermi s golden rule , _ i.e. _ @xmath10 where we have assumed an electric dipole transition with the electronic transition dipole moment ( @xmath11 ) , which is a function of normal coordinates of the initial electronic state .
the @xmath12-dimensional harmonic oscillator eigenstates of the initial and final electronic state are denoted by @xmath13 and @xmath14 with the corresponding harmonic energy vectors @xmath15 and @xmath16 , respectively .
the vibrational hamiltonians of the the initial and final electronic states are @xmath17 and @xmath18 , respectively .
@xmath19 is the boltzmann constant .
@xmath20 is the vibronic transition energy with respect to the @xmath9 transition energy .
the time - dependent part with the vibronic transition energy corresponds to a dirac delta distribution in frequency domain and accordingly this expression is usually evaluated in frequency domain , which is straightforward .
the spatial representation of the tcf in closed form can be found by evaluating the following quantum mechanical traces @xmath21 the traces can be evaluated with any complete basis . in our work coherent states
were used ( see _ e.g. _ refs .
@xcite ) with the duschinsky relation between initial and final state normal coordinates ( @xmath22 where @xmath23 and @xmath24 are the duschinsky rotation matrix and displacement vector , respectively and @xmath25 are the normal coordinates of the final state ) and the ht expansion of the electronic transition dipole moment ( @xmath26 where @xmath27 is the first derivative of @xmath28 with respect to @xmath29 . ) .
the vibronic spectral density function can be related to a probability density function ( pdf ) because the vibronic transition process follows a certain pdf .
if all cumulants or moments of a pdf are defined and available , the pdf can be reconstructed as follows @xmath30 where @xmath31 is the @xmath32-th order cumulant at temperature @xmath5 .
the cumulants of the spectral density function are normalised to give the total intensity @xmath33 .
moments ( cumulants and moments are inter - convertible ) can be obtained by partial derivatives of @xmath34 with respect to time , @xmath35 thus , cumulants can be evaluated analytically or numerically by evaluating partial derivatives of @xmath34 in eq . with respect to the time variable at @xmath36 .
analytic evaluation of the cumulants to arbitrary order within the linear ht approximation can be performed along the lines of the development in refs .
@xcite for the cumulants of fc profiles to arbitrary order . for numerical evaluation of low - order cumulants one needs to compute @xmath34 only at the first few time steps . to obtain the corresponding moments numerically , @xmath37 and @xmath38 are used for the calculation of even and odd moments , respectively , because @xmath39 in eq .
( see _ e.g. _ ref .
@xcite ) . the closed form of @xmath40 within the linear ht approximation can be found in ref .
therein , additional flexibility was introduced to the gf by distinguishing individual vibrational modes to allow for rigorous prescreening strategies and detailed control of the dynamics .
this can be done by assigning different time and temperature variables to each vibrational degrees of freedom .
the corresponding gf in an occupancy representation reads as follows @xmath41 \mathrm{e}^{-({\underline{v}}^{\mathrm{t}}{\mathbf{b}}{\underline{\epsilon } } + { \underline{v}}'^{\mathrm{t}}{\mathbf{b}}'{\underline{\epsilon } } ' ) } \ , , \label{eq : occfg}\end{aligned}\ ] ] where the general operators @xmath42 and @xmath43 , which can be products of momentum and position operators , are given instead of @xmath44 .
the time variables are assigned to the matrices @xmath45 = @xmath46 and @xmath47 = @xmath48 for initial and final vibrational modes , respectively .
different temperatures can be given to the initial and final vibrational degrees of freedom via @xmath49 = @xmath50 and @xmath51 = @xmath52 , respectively .
@xmath53 is the corresponding normalising factor .
the electronic @xmath0 transition of benzene is fc - forbidden in the electric dipole approximation ( @xmath54 ) such that only the ht terms contribute to the spectral density function .
the corresponding tcf is given as follows , here with same time and temperature ( @xmath4,@xmath5 ) for all vibrational degrees of freedom , @xmath55
a time increment of 0.51 fs and a grid with 2@xmath56 grid points are used for the corresponding fft calculations .
the experimental uv absorption spectrum of fischer @xcite is additionally shown in read .
right part of the figure : the dashed lines are drawn for the tcf - fft approach with a gaussian line shape function of with fwhm of 500 @xmath57 . a time increment of 0.10 fs and a grid with 2@xmath58 grid points are used for the corresponding fft calculations .
solid lines are drawn for the curve obtained by edgeworth expansion using up to 4-th order cumulants and , for 1000 k by edgeworth expansion using up to 8-th order cumulants .
, scaledwidth=45.0% ] the vibronic profiles for benzene s @xmath0 transition at zero kelvin and finite temperatures are calculated with two methods , namely tcf and time - independent cumulant expansion ( ce ) .
we use herein the term time - independent ce , which was employed in ref .
@xcite , to distinguish this ce from the conventional ( time - dependent ) ce ( see _ e.g. _ refs .
@xcite ) which involves time integration for the cumulant calculation .
the spectra obtained are displayed in fig .
[ fig : benzenespectrum ] . to compute the vibronic spectra via the tcf method
, the fftw library @xcite is used for fast fourier transform ( fft ) .
the approximate curves are generated for the ce with edgeworth expansion @xcite using the computed low - order cumulants .
some of the problems related to this type of expansion for the description of fc profiles are discussed in ref .
@xcite . the moments ( eq . [ eq : moments ] )
are calculated both analytically and numerically , the latter by taking partial derivatives of @xmath34 in eq .
[ eq : closedform ] with respect to time .
analytical and numerical results are compared in table [ table : moments ] . required input data from electronic structure calculations for benzene , _ i.e. _ molecular equilibrium structures and corresponding harmonic force fields for each electronic state ( @xmath59 and @xmath60 ) as well as first derivatives of the electronic transition dipole moments are taken from ref .
@xcite ( casscf / dzv ) .
these data have recently been compared to results obtained via analytical derivatives techniques for electronic transition dipole moments within a time - dependent density functional theory framework @xcite .
the vibronic structure methods employed in the present work are implemented in a development version of our vibronic structure program package hotfcht @xcite . the left hand side in fig .
[ fig : benzenespectrum ] shows vibronic profiles from tcf - fft which are convoluted by a lorentzian line shape function with full width at half maximum ( fwhm ) of 50 @xmath57 at temperatures elevating from 0 k to 1000 k. this fc - forbidden vibronic transition is mediated by the non - totally symmetric vibrational modes in the irreducible representation e@xmath61 .
the main feature of the vibronic spectrum is from progressions in the totally symmetric c - c stretching mode ( 963 @xmath57 ) building on the so - called false origin from a single excitation of a non - totally symmetric ( e@xmath61 ) in - plane bending mode ( 575 @xmath57 ) as indicated in the spectrum at zero kelvin . the calculated spectrum at 300 k is compared with the experimental data of fischer @xcite .
the two spectra agree fairly well in the low energy region but the computed peaks at higher energies are slightly shifted to larger wavenumbers due to the harmonic approximation . as temperature increases the vibrational structure becomes very congested and washed out . at 1000
k ( only employed for testing the method ) one can not see a resolved vibrational structure any longer , only the corresponding envelope . on the right hand side of fig .
[ fig : benzenespectrum ] the two methods ( tcf - fft and ce - edgeworth ) are compared for increasing temperatures .
the spectra are convoluted in the tcf - fft curves ( dashed lines ) with a gaussian line shape function of 500 @xmath57 for fwhm .
the second moments ( 4.51@xmath6210@xmath63 @xmath64 ) of the gaussian line shape function is added to the second moments of the vibronic spectrum to take the line shape function into account ( see ref .
@xcite for the rationalisation and details ) .
the relatively broad line shape function is used for the tcf - fft curves to have vibrationally relatively structureless spectra for comparison . at 0 , 300 and 500 k ,
the tcf - fft curves still show vibrational structure and the ce - edgeworth curves ( solid lines ) look like nonlinear regression curves of the corresponding tcf - fft spectra .
when the vibrational structures are also essentially smoothed out in the tcf - fft curves at 1000 k , the two approaches agree with each other extremely well . up to the 4-th order cumulants
are used for 0 , 300 , 500 k and up to 8-th order cumulants are computed for 1000 k. in table [ table : moments ] the moments computed numerically ( via derivatives ) and analytical are compared . at low orders and all temperatures
the two methods agree well and for higher orders still the agreement is satisfactory .
one of the advantages of the numerical method is that one only needs to compute the tcf for the first few time steps .
the analytical method usually meets a combinatorial problem in high order cumulant calculations due to the analytic derivatives of the inverse matrix .
the second advantage is that it is easy to include linear and nonlinear non - condon effects .
the third advantage is that one can incorporate general line shape functions which would not have well defined cumulants ( see the discussion on page 415 of ref .
@xcite ) .
mean excitation wavenumbers are computed for the ht active @xmath65 symmetric vibrational modes of the final state .
the mean excitation energy can serve as a parameter for the individual vibrational degrees of freedom as an effective reorganisation energy or a huang
( when normalised by its harmonic energy ) , which can be characterised as a function of structural deformation , frequency change , duschinsky mode coupling and temperature both in the condon and non - condon approximation . one might naively expect a larger mean excitation energy as the temperature increases but the mean values of high frequency modes ( 1665 and 3389 @xmath57 ) in some intermediate temperature rages are smaller than at zero kelvin .
this can be rationalised as follows : because the duschinsky mode mixing between low and high frequency modes is small in the present case , the high frequency modes can not obtain thermal energy from the low frequency modes efficiently .
thus the high frequency modes are almost thermally inactive , whereas the total intensity ( @xmath66 ) increases as temperature increases . in the mean energy calculation of the high frequency modes at finite temperatures
the denominator ( total intensity ) increases because low frequency modes accept thermal energy while the numerator ( excitation of high frequency modes ) stays constant .
therefore the mean excitation energies of high frequency modes are reduced at finite temperatures . if duschinsky rotation couples the low and high frequency modes significantly ( see _ e.g. _ ref .
@xcite ) , however , thermal energy can be transfered to the high frequency modes via the low frequency modes in the initial state , accordingly the mean excitation energies of high frequency modes can increase as temperature increases .
we have discussed a cumulant expansion method for describing non - condon transitions and applied it to the prototypical one - photon electric dipole @xmath0 transition of benzene , which is fc forbidden but ht allowed in the linear ht approximation .
the method is particularly powerful when one does not require all the details of the vibronic structures , but rather only quantities such as peak maximum , mean and variance of the spectral shape .
this method is computationally much cheaper than the sum - over - states and time - correlation function approach .
moreover , the information ( _ e.g. _ relevant energy window ) from the cumulant expansion method can be used in the calculation within the other two methods .
herein we compared a numerical approach for the calculation of cumulants with the results from an analytical scheme .
the results obtained numerically are still fairly good . with this method
, one can incorporate easily nonlinear non - condon terms and various line shape functions . in the time - correlation function calculation the real part and imaginary part at each time step
provide automatically the even and odd moments , respectively . in the first few time steps we already have the first few moments available and
the probability distribution function ( information ) becomes complete as time progresses .
financial support by the beilstein - institut , frankfurt / main , and computer time provided by the center for scientific computing ( csc ) frankfurt are gratefully acknowledged .
we are indebted to jason stuber for discussions .
26ifxundefined [ 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 * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop ( ) @noop * * ( ) @noop _ _ ( , , ) @noop * * , ( ) , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * ( ) @noop ( ) | when existing , cumulants can provide valuable information about a given distribution and can in principle be used to either fully reconstruct or approximate the parent distribution function . a previously reported cumulant expansion approach for franck - condon profiles [ faraday discuss .
, 150 , 363 ( 2011 ) ] is extended to describe also the profiles of vibronic transitions that are weakly allowed or forbidden in the franck - condon approximation ( non - condon profiles ) . in the harmonic approximation the cumulants of the vibronic spectral profile can be evaluated analytically and numerically with a coherent state - based generating function that accounts for the duschinsky effect . as illustration ,
the one - photon @xmath0 uv absorption spectrum of benzene in the electric dipole and ( linear ) herzberg - teller approximation is presented herein for zero kelvin and finite temperatures . |
emission observed from black hole binaries ( bhbs ) is variable on both long and short timescales . on the longest timescales , changes in mass accretion rate drive changes in spectral state . at the lowest luminosities ,
the system is typically seen in the low / hard state , where the spectrum is dominated by a hard ( photon index @xmath1 ) power law tail , peaking at 100 kev .
as the source brightens , there is an increasing contribution from the disc at low energies and the tail softens towards @xmath2 ( intermediate state ) .
the tail can then either remain strong as the source flux increases , giving the very high state ( @xmath3 ) or it can carry only a very small fraction of the luminosity , giving the high / soft state ( @xmath4 ) ( e.g. see reviews by remillard & mcclintock 2006 ; done , gierlinski & kubota 2007 , hereafter dgk07 ) . much of this long term spectral evolution can be successfully described by the truncated disc model ( dgk07 ) . here , the assumption is that the cool , geometrically thin , optically thick disc is truncated at some radius , @xmath5 , which is much larger than the last stable orbit , @xmath6 at the lowest luminosities .
the inner accretion flow extends from @xmath5 to an innermost radius , @xmath7 , forming some sort of hot , geometrically thick , optically thin ( optical depth @xmath8 ) flow , similar to an advection dominated accretion flow ( adaf , narayan & yi 1995 ) .
decreasing the truncation radius of the thin disc with increasing luminosity ( and therefore average mass accretion rate ) leads to a stronger disc component and greater illumination of the hot flow by the cool disc photons and hence to a softer comptonised tail .
this spectral evolution has a natural end when @xmath9 , marking the transition to the disc dominated soft states seen at higher luminosities .
there is also rapid variability on timescales of 100 - 0.1s which is almost exclusively linked to the comptonised tail ( e.g. churazov et al 2001 ) .
this is as expected as the disc emission can only vary on a viscous timescale of @xmath10 s i.e. 500 s at the last stable orbit for fiducial thin disc parameters of @xmath11 and @xmath12 .
a power spectral analysis of the rapid variability shows that it is generally composed of a broadband noise continuum , with a low frequency quasi - periodic oscillation ( lf qpo ) superimposed on this .
the broadband noise can be very roughly characterized as a twice broken power law , with @xmath13 at the lowest frequencies , breaking to @xmath14 above the low frequency break @xmath15 , and then breaking again to @xmath16 above a high frequency break at @xmath17 .
the low frequency break is tightly correlated with the spectral state of the source , with @xmath18 hz for the dimmest low / hard states , and increasing to @xmath19 hz as the source softens through the bright low / hard states into the intermediate state .
the lf qpo properties are also tightly correlated with this power spectral and energy spectral evolution , with qpo frequency @xmath20 increasing from 0.1 - 10 hz , as the coherence , power and harmonic content of this feature also increase ( e.g. van der klis et al 2004 ) .
again , this can be interpreted qualitatively in the truncated disc picture if the low frequency break and qpo are somehow associated with the truncation radius .
as the disc extends further inwards , the spectrum softens , and all frequencies associated with this radius increase ( e.g. dgk07 ) .
however , to go beyond this into a quantitative description requires a specific model for both the broadband noise and qpo .
this is problematic , despite both having been known for many decades ( e.g. van der klis 1989 ) .
there are multiple potential models for the lf qpo in the literature which fall into 2 main categories : those associated with a geometrical misalignment of the accretion flow and black hole spin ( stella & vietri 1998 ; fragile , mathews & wilson 2001 ; schnittman 2005 ; schnittman et al 2006 ; ingram , done & fragile 2009 , hereafter idf09 ) , and those associated with wave modes of the accretion flow ( wagoner et al 2001 ; titarchuk & oscherovich 1999 ; cabanac et al 2008 ) .
most of these concentrate on matching the qpo frequency , but the spectrum of the qpo gives additional constraints .
this is similar to that of the spectrum of the broadband variability , showing that they both arise predominantly from the comptonising region rather than the disc ( e.g gilfanov et al 2003 ; sobolewska & zycki 2006 ) , favoring models in which the modulation arises directly from the comptonised emission e.g. idf09 , where the qpo is set by lense - thirring ( vertical ) precession of the entire hot inner flow interior to the disc truncation radius at @xmath5 , or by a mode of the hot inner flow ( cabernac et al 2008 ) .
the physical origin for the viscosity of the flow is the magneto - rotational instability ( mri : balbus & hawley 1991 ) .
this is inherently variable , with large fluctuations in all quantities , both spatially and temporally ( krolik & hawley 2002 ) , making it a natural origin for the broad band noise ( noble & krolik 2009 ; ingram & done 2010 ; dexter & fragile 2011 ) .
however , these fluctuations also effectively shred any coherent wave modes in the flow ( see e.g. reynolds & miller 2009 ; henisey et al 2009 ) .
this effectively rules out trapped wave propagation as the origin of the lf qpo , leaving lense - thirring precession as the most likely candidate .
precession of the entire hot flow from @xmath5 to @xmath7 can match the observed lf qpo frequency in both bhbs ( idf09 ) _ and _ neutron stars ( ns ; ingram & done 2010 ) , and provides a clear mechanism to match the spectrum as this is a modulation of the comptonising region .
thus the entire power spectrum can be explained by mri fluctuations in a hot flow , which is also precessing around the black hole .
however , the power spectrum does not represent all the information contained in the variability , as it uses only fourier amplitudes , not phases .
this is important as the light curves contain additional correlations which give a linear rms - flux relation ( uttley & mchardy 2001 ) .
short segments of a longer light curve have a mean , @xmath21 , and variance , @xmath22 , which are related such that @xmath23 ( after binning : uttley & mchardy 2001 ) .
this is equivalent to the flux on these timescales having a log - normal distribution ( negoro et al 2000 ) , and rules out simple models of the variability where the light curve is made from adding together multiple , uncorrelated events ( uttley & mchardy 2001 ; see also dgk07 ) .
instead , this can be produced if the light curve is made from a multiplicative process , rather than an additive one .
again , the mri in the hot flow gives a physical interpretation to this .
the mri at large radii produces intrinsic fluctuations in the density of the flow .
these fluctuations propagate down to smaller radii on a viscous timescale , so all higher frequency fluctuations are smoothed out .
these smoothed and lagged fluctuations in mass accretion rate modulate the mri fluctuations produced by the next radius , and so on , down to the smallest radii in the flow ( luybarski 1997 : hereafter l97 ) .
this naturally produces a light curve which has an rms - flux relation / log - normal flux distribution ( kotov et al 2001 : hereafter k01 ; arevalo & uttley 2006 : hereafter au06 ; also see misra & zdziarski 2008 ) .
here we use these ideas to build a model for the entire power spectrum , where the broadband noise arises from propagation of mri fluctuations through the hot flow from @xmath5 to @xmath7 and the lf qpo arises from lense - thirring precession _ of the same hot flow_. we fit this model to a sequence of power spectra of xte j1550 - 564 observed during a transition from low / hard to intermediate state , and show that these give a good _ quantitative _ fit for @xmath5 moving from @xmath24 , as also required to explain the sequence in energy spectra during this transition .
this gives very strong support to the underlying truncated disc geometry , and for lense - thirring precession as the physical origin for the lf qpo .
we first introduce the simplified toy model considered in idf09 and ingram & done ( 2010 ) , whereby the low frequency break , @xmath15 , occurs at the viscous frequency of the truncation radius and the qpo frequency , @xmath25 , is the precession frequency of the flow . using the shakura sunyaev ( 1973 ) viscosity prescription , @xmath26 where @xmath27 is the viscosity parameter , @xmath28 the flow semi - thickness , @xmath29 the keplerian frequency and @xmath30 is the infall velocity .
the lense - thirring precession frequency is then a weighted average of the point particle precession frequency at each radius in the hot flow , so @xmath31 ( liu & melia 2002 ) where @xmath7 is the innermost point of the flow ( i.e. the surface density is negligible interior to this ) , @xmath32 is the surface density and @xmath33\ ] ] is the point particle lense - thirring precession frequency for a dimensionless spin parameter @xmath34 ( merloni et al 1999 ) . here
, @xmath35 is dimensionless , expressed in units of @xmath36 . solving this assuming a power law form for surface density in the hot flow @xmath37 between an inner and outer radius for the hot flow @xmath38 and @xmath5 ( fragile et al 2007 , idf09 ) gives @xmath39 } { r_o^{5/2-\zeta}r_i^{1/2+\zeta } [ 1-(r_i / r_o)^{5/2-\zeta}]}\frac{c}{r_g}.\ ] ] hence this model predicts the relation between @xmath25 and @xmath15 which can be compared to the multiple observations of these frequencies in both black holes and neutron stars ( e.g. wijnands & van der klis 1999 ; klein - wolt & van der klis 2008 ) .
the observed relation is continuous , implying that these frequencies show the same behavior in both sources , i.e. that neither can depend strongly on any property of the neutron star surface but are instead set by the accretion flow itself .
we re - plot this data in figure [ fig : wkplot ] , normalizing the frequencies by mass for a fiducial mass of @xmath40 and @xmath41 for neutron stars ( red triangles ) and black holes ( black circles ) , respectively .
this shows even more clearly that the two different types of object show the same observed relation between these frequencies as they now occupy the same range .
the blue lines show the prediction of the toy model , where the hot flow has constant @xmath42 and @xmath43 , surface density constant with radius ( i.e. @xmath44 : fragile et al 2007 ) between @xmath5 and @xmath38 , where @xmath38 is given by the bending wave radius .
warps in a large scale height flow are communicated via bending waves which have wavelength @xmath45 and so are smooth at large @xmath35 and oscillatory at small @xmath35 .
the bending wave radius ( @xmath46 ; fragile et al 2007 ; 2009 ; fragile 2009 ; idf09 ) marks the transition between the two regimes .
we show @xmath47 ( dot - dashed ) , @xmath48 ( dashed ) and @xmath49 ( solid ) .
while this very simple model predicts frequencies which are fairly close to the observations , it is clear that the gradient of this model in log space is different from that observed .
plainly the assumptions above are very simplistic .
global analytic models of the hot flow with a standard @xmath27 viscosity do not have @xmath50 as they depart from the self - similar solutions at @xmath51 due to the requirement that the flow becomes supersonic ( narayan , kato & honma 1997 ; gammie & popham 1998 ) .
full numerical simulations also show that @xmath27 is not constant ( e.g. fragile et al 2007 ; 2009 ) .
ingram & done ( 2010 ) also suggest that @xmath52 can change in neutron stars as the material piles up onto a boundary layer . however , the similarity between the mass scaled frequencies seen in neutron stars and black holes shown in fig .
[ fig : wkplot ] make this now seem unlikely to be an important effect as it would not affect the black holes . here
then we simply assume that @xmath53 is a power law function of radius , so that @xmath54 .
we choose values for @xmath55 and @xmath56 which allow us to match the data in figure [ fig : wkplot ] .
we see good agreement with the observations for @xmath57 and @xmath58 ( green lines ) , again for @xmath47 ( dot - dashed ) , @xmath48 ( dashed ) and @xmath49 ( solid ) .
we use this specific prescription for the viscous frequency in the following section .
we consider a model where local fluctuations in the mass accretion rate of the flow propagate down towards the central object ( e.g. l97 ; k01 ) .
our method mainly follows that of au06 , with a few small differences .
we split the flow into annuli , characterized by a radius @xmath59 and width @xmath60 , with logarithmic spacing so @xmath61 is a constant for all annuli from @xmath5 to @xmath38 . as @xmath62 in our prescription , constant @xmath61 also implies constant @xmath63 which au06 show is the feature required to produce a linear sigma - flux relation .
mri fluctuations throughout the flow generate variability in all quantities .
fluctuations in mass accretion rate will most likely be damped by the response of the flow on timescales shorter than the local viscous timescale .
we therefore assume that the generated power spectrum of mass accretion rate fluctuations at radius @xmath59 is given by a zero centred lorentzian cutting off at the viscous frequency @xmath64 where @xmath65 as discussed at the end of section [ sec : mod ] and a tilde denotes a fourier transform .
we start at the outermost annulus , so @xmath66 , and generate the time dependent fluctuations in mass accretion rate , @xmath67 , from equation [ eqn : poff ] using the method of timmer & koenig ( 1995 ) .
we normalize each @xmath68 to have a mean of unity and fractional variability @xmath69 where @xmath70 is the fractional variability per decade in radial extent and @xmath71 is the number of annuli per decade in radial extent .
thus the mass accretion rate across the first annulus is @xmath72 where @xmath73 is the mean mass accretion rate .
this then propagates inward to the second annulus , travelling a distance @xmath74 , which takes a time @xmath75 .
when it arrives at @xmath76 , it has been filtered by the response of the flow which we take from psaltis & norman ( 2000 ) to get @xmath77 ^ 2 } } .
\label{eqn : pandn}\ ] ] the mass accretion rate at the @xmath78 annulus is then given by @xmath79 where @xmath80 .
however , equation [ eqn : pandn ] only filters out fluctuations on much shorter timescales ( by a factor @xmath81 ) than the typical timescales generated in the annulus ( equation [ eqn : poff ] ) and so we can say @xmath82 to a very good approximation . the mass accretion rate at the @xmath78 annulus is therefore given by @xmath83 until the @xmath84 annulus which is @xmath38 . to transform this into a light curve
requires an emissivity , @xmath85 such that the luminosity from each annulus is given by @xmath86 where we assume the emissivity @xmath87 where @xmath88 is a boundary condition . for a newtonian thin disc , @xmath89
and we have the stress free inner boundary @xmath90 but we note that the large scale magnetic fields present in the large scale height flow can give a stressed inner boundary condition @xmath91 ( agol & krolik 2000 ; beckwith hawley & krolik 2008 ) .
we also allow @xmath92 to be a free parameter as the emission need not exactly follow the radial dependence of gravitational energy release as long as the _ total _ energy release is gravitational .
a different emissivity for different energy bands gives a way for the model to predict frequency dependent time lags between hard and soft x - ray bands ( k01 ; au06 ) .
this will be the subject of a subsequent paper .
@xmath93 figure [ fig : broadband]a shows 20 s of the resulting light curve for a fiducial set of input parameters for a black hole mass of @xmath94 and a spin of @xmath48 .
we assume @xmath95 , @xmath96 , @xmath97 , @xmath57 , @xmath58 , @xmath98 with a stressed inner boundary condition ( see section [ sec : mod ] ) . we calculate the light curve with @xmath99 time points , corresponding to @xmath100 s ( a typical length for an rxte observation ) of data on a time binning of @xmath101 s , and 30 radial bins .
figure [ fig : broadband]b ( red ) shows the psd of this light curve , while the black and green points show the effect of changing the number of radial bins to 10 and 100 , respectively .
clearly , the high frequency power is not well resolved with only 10 radial bins , while the difference between 30 and 100 is very small . hence
in all that follows , we use 30 radial bins for each simulation .
the psd shows the same characteristic broadband noise features as are seen in the power spectra of black hole binaries , namely band limited noise , with low and high frequency breaks , peaking between 0.1 - 10 hz .
figure [ fig : broadband]c shows the rms - flux relation for the fiducial light curve , derived from splitting this into 4s segments . as with the data ( uttley & mchardy 2001 ) , we see a large scatter before binning ( gray points ) but , after binning ( red crosses ) , we retrieve a linear flux - rms relation ( au06 ) .
the major prediction of the truncated disc / hot inner flow model is that the spectral softening as the source brightens from a low / hard through to intermediate states is caused by the truncation radius of the thin disc moving inwards ( e.g. dgk07 ; gierlinski , done & page 2008 ) .
this radius also sets the outer edge of the hot flow , so this predicts that @xmath5 decreases also .
figure [ fig : fvar ] shows the predicted psd for @xmath102 , @xmath103 and @xmath104 , as required to match the energy spectral evolution ( and low frequency qpo : idf09 ) , with all other parameters held constant at the fiducial model values described above .
the model _ predicts _ that decreasing the outer radius of the hot flow leads to less low frequency power , while the high frequency power remains constant .
this is precisely what is seen in the psd of the data ( dgk07 ; gierlinski , nikolajuk & czerny 2008 ) .
this is the first physical model of the power spectral behavior which naturally reproduces the observations .
the low frequency break is close to the frequency of the viscous timescale at @xmath5 , as proposed by e.g. by churazov et al ( 2001 ) ; gilfanov & areief ( 2005 ) ; dgk07 ; ingram & done ( 2010 ) .
however , the high frequency break is _ not _ at the viscous frequency at @xmath38 .
we explore the origin of the high frequency break below .
= 6.5 cm @xmath105 figure [ fig : cor]a shows a model where the variability at each radius is a lorentzian at the local viscous frequency ( see equation [ eqn : poff ] ) , but with no propagation so there is no causal connection between annuli .
we show the psd of the resulting @xmath68 functions from 5 of the 30 individual radial annuli , from @xmath5 to @xmath38 as the gray lines on figure [ fig : cor]a .
these peak , as expected , at @xmath106 and @xmath107 .
the total variability ( red ) is an emissivity weighted sum of these fluctuations , but since they are uncorrelated , the effect of this is to strongly dilute the total variability seen .
this total psd does have @xmath108hz as our emissivity weighting strongly favors the smallest radii , but @xmath109 ( @xmath110hz and @xmath111hz respectively ) .
in fact , to achieve @xmath14 as observed , we would have to assume a completely flat emissivity profile , which seems very unlikely .
more fundamentally , such uncorrelated fluctuations _ can not _ reproduce a linear rms - flux relation .
this is in sharp contrast to a model where fluctuations propagate down in radius .
the resulting psd from the same set of radii are shown in figure [ fig : cor]b , where the power in each annulus increases strongly with radius as the mri power generated in each annulus is modulated by the propagating fluctuations from all radii prior to it .
the red line shows the resulting emissivity weighted power spectrum from the total flow assuming that time lags between radii are negligible .
this preserves the maximum correlation between variability at different annuli i.e. gives the least dilution between fluctuations in different annuli .
this is very different to that in figure [ fig : cor]a , both in normalization and shape .
the normalization is dramatically enhanced because the long timescale fluctuations are correlated together , so at low frequencies the power from different radii add together as they are in phase .
this gives @xmath112hz as the correlated variability weighting to larger radii is stronger than the emissivity weighting to smaller radii .
however , at the fastest timescales , the power is mainly generated at the smallest radii , so it does not correlate with any other fluctuations generated at larger radii , so is not enhanced in the same way .
the blue line shows how time delays dramatically change the high frequency break as the propagation time prevents the mass accretion rate from two consecutive annuli from being correlated on time scales shorter than @xmath113 .
this reduces the correlation between the fastest timescale variability , strongly suppressing high frequency power .
thus in the propagating fluctuation model , the low frequency break is @xmath114 but @xmath115 .
= 6.5 cm we use an emissivity to translate the fluctuations in mass accretion rate to a luminosity .
this emissivity is in two parts , firstly a power law dependence in radius , and secondly a boundary condition .
our fiducial model parameters have @xmath98 and a stressed boundary condition , @xmath91 .
this emissivity peaks at @xmath38 , so fluctuations from the very smallest radii are given most weight .
figure [ fig : boundary ] compares this ( black line ) with results using the same power law radial dependence , but with a stress
free inner boundary condition ( red line ) , @xmath116 .
this emissivity goes to zero at the innermost radius , so the highest frequency fluctuations are strongly suppressed .
however , this also has a more subtle effect on the region between the two breaks , as there is a gradual decrease in weighting of fluctuations below @xmath117 , and a stronger weighting to the fluctuations at larger radii , giving the tilt between @xmath15 and @xmath17 .
this effect is similar to that of changing the radial dependence of the emissivity .
the green line shows @xmath89 with a stress - free boundary condition , showing an even stronger tilt to the psd between @xmath15 and @xmath17 ( green ) .
however , it is also similar to changing the inner radius of the flow .
the blue line in figure [ fig : boundary ] shows the resulting psd from @xmath89 and a stress - free inner boundary condition with @xmath118 .
thus there are degeneracies between the two parts to the emissivity and the inner radius , making it unlikely that they can all be uniquely constrained by the observed psd .
it is clear from this analysis that while the low frequency break is fairly strongly linked to the viscous timescale of the outer radius of the hot flow ( as assumed in section [ sec : mod ] ) , the high frequency break is rather more complex , depending on propagation correlations , emissivity , boundary condition and inner radius in addition to the viscous timescale .
this makes it difficult to directly associate the high frequency break with any physical parameter of the models .
instead , we now use the additional information from the qpo to remove some of the degeneracies inherent in this model for the broadband noise .
[ sec : qpo ] for our fiducial model , we used the observed relation between the low frequency break and lf qpo to set the radial dependence of the viscous timescale , assuming that the low frequency break was set by the viscous timescale at @xmath5 and that the qpo was lense - thirring precession of the entire hot flow from @xmath5 to @xmath38 ( section [ sec : mod ] ) .
this assumed that the surface density of the hot flow , @xmath119 between @xmath5 and @xmath38 , with @xmath44 .
however , the broadband noise model described above _ calculates _ a self - consistent surface density as mass conservation implies @xmath120 ( frank , king & raine 1992 ) where @xmath30 is now expressed in units of @xmath121 , @xmath122 in units of @xmath73 , @xmath32 in units of @xmath123 and @xmath35 in units of @xmath124 . using our velocity prescription
, we can then easily show @xmath125 this means that , for the time averaged surface density , @xmath126 giving extra physical motivation for the parameters used in section [ sec : mod ] ( @xmath44 and @xmath127 ) .
figure [ fig : surfaced ] shows @xmath128 plotted at a number of different times ( @xmath129 ) along with the corresponding @xmath130 function .
@xmath130 is quite clearly more variable at small @xmath35 .
this is because we have assumed the variability _ generated _ in each logarithmic annulus to be the same but annuli at smaller radii include also the fluctuations that have propagated down from large @xmath35 and so the _ emitted _ variability is greater ( see figure [ fig : cor]b ) .
we do not see a drop off in surface density at the bending wave radius like that seen in simulations ( e.g. fragile 2009 ) because we assume that the infall velocity can be given by a power law .
it is clear that , for the surface density drop off at a given point , the infall velocity must accelerate at that point . in a future paper
, we will investigate this model with a more advanced velocity prescription .
@xmath131 therefore the broadband noise model above , set by @xmath38 , @xmath5 and @xmath70 _ predicts _ the qpo frequency at any point in time .
the fluctuations in surface density with time _ predict _ that the qpo frequency changes , i.e. it is quasi
periodic rather than truly periodic .
however , the precession frequency will not respond instantaneously to these changes , as their effect is only communicated across the entire hot flow by bending waves .
these travel at the sound speed , faster by a factor @xmath132 than the viscous timescale across the region , so we calculate the qpo frequency every @xmath133 s rather than at every point .
we then average these values to get the predicted qpo frequency , @xmath25 , and use the dispersion around this to set the _
_ variance of these qpo frequencies , @xmath134 .
[ fig : qvsqpo ] shows @xmath135 as a function of @xmath25 as @xmath5 varies from 30010 in the fiducial model . this decrease in @xmath5 not only leads to an increase in qpo frequency , but also to a decrease in the qpo width , or equivalently , an increase in its coherence / quality factor @xmath136 .
this correlation is well known in qpo data from bhb ( e.g. belloni , psaltis & van der klis 2002 ; rao et al 2010 ) .
our model provides the first physical explanation for this effect as the smaller the radial extent , the higher the qpo frequency , but also the smaller the fluctuation power , giving smaller jitter in frequency .
the red squares in fig [ fig : qvsqpo ] show the _ observed _ frequency and width of the qpo from data from the 1998 rise to outburst of the bhb xte j1550 - 564 ( see section 7 ) .
the model matches the trend in the data fairly well , and forms a lower limit to the width of the qpo
. however , other effects such as the on - time of the qpo ( see lachowicz & done 2010 ) can decrease the coherence of the signal , so here we simply use the data to determine the qpo width ( parametrized as a quality factor @xmath136 ) after using the broadband noise surface density to determine the qpo frequency , @xmath25 .
= 6.5 cm @xmath137 = 6.5 cm the model also predicts another correlation , one between the qpo frequency and flux on short timescales .
the top plot of figure [ fig : lvsqpo ] shows this for the fiducial model ( i.e. @xmath96 ) , with precession frequency calculated every @xmath138s together with the instantaneous luminosity at that time .
after binning ( red crosses ) , there is a clear linear relation between the two .
this happens because both the qpo frequency and the luminosity depend on the mass accretion rate fluctuations .
a perturbation in mass accretion rate at large @xmath35 will lead to a perturbation in the surface density .
this will reduce the precession frequency but will have little effect on the luminosity because the emissivity is quite steeply weighted towards small @xmath35 . later on ,
this perturbation will have propagated inwards to small @xmath35 where it has the effect of increasing the precession frequency , but now also has much more of an effect on the luminosity .
heil et al ( 2011 ) have recently discovered this correlation in data from the 1998 rise to outburst of xte j1550 - 564 ( the same data as we will be considering in section [ sec : data ] ) .
they also find that the gradient of this relation is steeper for observations with a higher qpo frequency .
this is also predicted by the model as illustrated in the bottom plot of figure [ fig : lvsqpo ] where we have measured the gradient of the @xmath25-@xmath139 relation and the average qpo frequency for @xmath140 different @xmath5 values .
there is clearly a very strong correlation as is seen in the data .
this is because an _ absolute _ change in precession frequency depends on a _ fractional _ change in mass accretion rate whereas an _
absolute _ change in luminosity depends on an _ absolute _ change in mass accretion rate . the same absolute change in mass accretion rate at a given radius and time constitutes a larger _ fractional _ change for small @xmath5 than for high @xmath5 .
therefore the luminosity will experience exactly the same change in both instances but the precession frequency will undergo a larger change when @xmath5 is smaller . the fact that these are _ predicted _ properties of the model constitutes strong support for its validity . in the following section
, we will fit the psd from our model to data . for this
, we will need to predict a _ shape _ for the qpo light curve as well as a frequency .
the data show that the qpo has a power spectrum which can be represented by a lorentzian at the fundamental frequency , @xmath25 , together with its second and third harmonic and sub - harmonic i.e. at @xmath141 , @xmath142 and @xmath143 ( e.g. belloni , psaltis & van der klis 2002 ) .
our model for the qpo in terms of lense - thirring precession _ predicts _ the shape of the modulation of the emission from the hot flow via variation of projected area , self - occultation and seed photons ( idf09 ) .
we will explore this further in a later paper ( ingram , done & @xmath144ycki in prep ) , but here we simply assume that all the harmonics have the same quality factor , @xmath145 , and allow the power in each harmonic to be a free parameter .
we then generate a qpo light curve , @xmath146 , using these narrow lorenzians as input to the timmer & koenig ( 1995 ) algorithm , and add this to the light curve already created for the broadband noise .
we show an example of the final predicted psd in figure [ fig : qpoandbb ] , using the fiducial model parameters with @xmath102 ( black ) , @xmath103 ( red ) and @xmath104 ( green ) .
for clarity , we have set the normalization of all the harmonics other than the fundamental to @xmath147 , set the width of the qpo using the model prediction of @xmath135 , and set its rms power @xmath148 .
these psd show all the main features seen in the data during spectral transitions of bhb ( gierlinski , nikolajuk & czerny 2008 ) .
we now have a model which can produce both the broad band noise and qpo self consistently from the same changing geometry as required by the corresponding evolution of the energy spectra , namely a changing outer radius of the hot flow ( set by the changing inner radius of the thin disc ) .
this model gives the major correlation between the low frequency break and low frequency qpo .
the propagating fluctuations naturally give the rms - flux relation , while the qpo model also gives a framework in which to understand the increase in coherence of the qpo with frequency , and even the short timescale correlation of the qpo frequency with flux .
plainly , the next step is to use this model to fit real data .
we use rxte data from the 1998 outburst of xte j1550 - 564 ( remillard et al 2002 ; sobczak et al 2000 ; rao et al 2010 ; wilson & done 2001 ) .
we look at 5 specific observations with observational ids : 30188 - 06 - 03 - 00 , 30188 - 06 - 01 - 00 , 30188 - 06 - 01 - 03 , 30188 - 06 - 05 - 00 and 30188 - 06 - 11 - 00 ; hereafter observations 1 - 5 respectively .
we only consider energy channels 36 - 71 ( corresponding to 10 - 20 kev ) in order to avoid disc contamination . in all our results
so far we have shown power spectra produced by averaging the logarithm of the periodogram as papadakis & lawrence ( 1993 ) show that , for a red noise variability process , the error estimate converges to gaussian with only @xmath149 samples , rather than the @xmath150 required for averaging the linear power estimates .
our simulations average over @xmath151 samples of the psd but the number of segments from the real data can not be chosen as they are simply set by the length of the available data .
these give @xmath152 and @xmath153 respectively , implying that the errors on the psd from the data are only approximately gaussian .
a further complication is that the error distribution of the periodogram is dependent on the shape of the underlying power spectrum ( e.g. mueller , madejski & done 2004 ; mueller & madejski 2009 ) meaning that even @xmath154 may not be enough to give gaussian errors in this particular case .
nonetheless , we choose to use @xmath155 minimization for simplicity , but we then do _ a posteriore _ checks on the goodness of fit using the rejection probability method of uttley et al ( 2002 ) and markewicz et al ( 2003 ) .
this entails first calculating @xmath156)^2}{(\overline{\delta p_{sim}(f)})^2}\ ] ] where @xmath157 is the logarithmically smoothed simulated periodogram , with error @xmath158 , to be compared to @xmath159 raw observed periodograms , @xmath160 .
we then re - simulate many ( 512 ) light curves and calculate many different values of @xmath161)^2}{(\overline{\delta p_{sim}(f)})^2}.\ ] ] the rejection probability , @xmath162 , is given by the percentile of @xmath163 values which are lower than the one @xmath164 value .
for example , if 128 of the @xmath163 values were smaller than the one @xmath164 value , the rejection probability is @xmath165 .
this means there is a @xmath165 chance that the the observed light curve was generated by a different underlying variability process than the model .
one aspect of using real data is that there are uncertainties associated with each point on the light curve .
since we use the logarithmically averaged psd we need the power to be positive definite so we can not do white noise subtraction of the data psd .
instead we add white noise to the simulation , and compare this with the total ( including noise ) power spectrum of the real data ( see also uttley et al 2002 ) .
we incorporate our model for the power spectrum into xspec , using the local model functionality .
our model outputs the logarithmically averaged total ( white noise included ) periodogram as a function of fourier frequency rather than the more familiar flux as a function of energy .
we then fit this to the similarly logarithmically averaged total power spectra from the real data .
the model parameters are the physical parameters governing the behavior of the flow .
the most fundamental of these is the relation governing the viscous frequency at any radius .
we have parametrized this as a power law function of the keplarian frequency at each radius , @xmath166 , but the parameters @xmath55 and @xmath56 are unknown , so are free in the fits , as is @xmath70 , the intrinsic level of the mri fluctuations in mass accretion rate .
then there are the geometric parameters @xmath5 and @xmath38 , where @xmath38 also couples to the emissivity , both in terms of the index @xmath92 and the boundary condition ( either stressed or stress free ) .
the combination of @xmath5 , @xmath38 and the surface density profile fixes the qpo frequency , but we leave its width ( parametrized as a quality factor , q ) and normalization as free parameters .
we also allow the normalization of the second , third and subharmonic to be free parameters , but fix these to have the same q as that of the fundamental .
the final free parameter is the normalization of the simulated white noise level , @xmath167 , which we then subtract from both data and model in order to plot results .
the truncated disc model predicts that @xmath5 is the major parameter which changes across this transition , where @xmath38 remains fixed .
hence we fit all 5 power spectra simultaneously in xspec , with @xmath38 tied across all 5 datasets . because of the correlation between @xmath38 and both parts of the emissivity ( see section [ sec : boundary ] ) , we choose to compare results with some of these parameters fixed at different values .
@xmath168 @xmath169 in this section we fit our model to the @xmath170 observations considered assuming a stressed inner boundary condition . figure [ fig : r - gammas ] displays the best fit models and data with the inner radius and emissivity index required to be constant across the @xmath170 observations .
these appear by eye to be good fits , and this is confirmed by their rejection probabilities of @xmath171 , @xmath172 , @xmath173 , @xmath174 and @xmath175 for observations 15 , respectively .
we find the inner radius to be @xmath176 which , although low , is greater than the horizon of the black hole ( @xmath177 for our chosen @xmath48 ) .
the truncation radius moves from @xmath178 .
however , the fit requires that the power law index governing the viscous frequency , @xmath56 , increases across the observations , from @xmath179 .
we will discuss the significance of this in section [ sec : para ] . in the previous section , we recovered an inner radius fairly close to the horizon ( @xmath176 ) for fits with a stressed boundary condition .
this assumption implies a causal connection to material closer in which , given the proximity of the emission to the horizon , may not be physical .
hence the stress free inner boundary condition may be more physically appropriate .
we repeat the fits assuming this , with results shown in figure [ fig : r - gammasf ] .
we recover an even lower inner radius ( @xmath180 ) , and similar values for the truncation radius ( @xmath181 ) and power law index governing the viscous frequency ( @xmath182 ) . by eye
, these again appear to be excellent fits ; in fact they look nearly identical to figure [ fig : r - gammas ] and have only a marginally higher @xmath155 value .
however , these have rejection probabilities of @xmath183 , @xmath183 , @xmath184 , @xmath185 and @xmath186 for observations @xmath187 respectively , so they are all statistically unacceptable .
a rejection probability of @xmath183 means that all of the 512 simulated psds had @xmath163 smaller than the one ` observed ' @xmath164 , making it extremely unlikely that the observed data is part of the simulated distribution .
this discrepancy between @xmath155 and rejection probability points to the complex nature of the statistics of the periodogram .
we discuss this in more detail in the appendix , but here we simply note that these results do _ not necessarily _ mean that there is no good fit for the stress free version of the model : they just mean that the real best fit parameters do not give the lowest @xmath155 .
the previous fits with a stressed boundary condition also may not be the best fit in terms of rejection probability , the difference is that with the stressed boundary condition the minimum @xmath155 solution was also a statistically acceptable fit on rejection probability .
we will explore this in more detail in a future paper ( ingram & done , in preparation ) however , as the fits appear to be good by eye , it seems very likely that there should be a set of parameters very close to the ones that minimize @xmath155 that give much lower rejection probabilities .
although the statistical complexities of the periodogram currently limit the model fitting to data , the parameters we derive are still most likely very close to the actual best fit parameters , and all four fits in the previous section came out with similar results for @xmath5 and the viscous frequency radial dependence . while the change in @xmath5 is predicted from the truncated disc / hot inner flow geometry proposed to model the correlated spectral changes , the change in viscous frequency radial dependence is not .
there are two ( at least ) scenarios in which this could occur .
firstly , this could indicate that the viscous frequency is well approximated by a power law but that the form of this changes as the truncation radius moves inwards ( as assumed in ingram & done 2010 for the neutron star power spectra ) .
there is currently no theoretical model in which this occurs .
a more attractive ( to us ) possibility is that a power law is a poor approximation for the viscous frequency and , as the truncation radius moves , the power law that best approximates the true shape of the viscous frequency with radius changes .
this would imply that there is a stationary function @xmath188 ( and by extension @xmath189 ) which would allow this model to fit all 5 observations with only a change in @xmath5 .
we note that exactly this behavior is predicted in global simulations of an adaf as the flow has to accelerate strongly to pass through a sonic point ( e.g. narayan , kato & honma 1997 ; gammie & popham 1998 ) .
these models give an in - fall velocity @xmath189 which is well approximated by a quadratic in log space ( see fig 1 in gammie & popham 1998 ) .
this will also allow the model to display a drop off in surface density in agreement with simulations as discussed in section [ sec : qpo ] .
approximating such a velocity law by a power law of index @xmath56 will result in @xmath56 increasing as the radial extent of the flow decreases , similar to the observed trend .
we will explicitly fit such models in a later paper ( ingram & done in preparation ) .
the truncated disc / hot inner flow model designed to describe the spectral evolution of bhb can also give a self consistent geometry in which to model the correlated evolution of the power spectrum .
propagating fluctuations through a hot flow which extends from and outer to inner radius , @xmath190 , can produce the band limited noise characteristic of the continuum power spectrum , as well as producing the rms - flux relation ( l97 ; k01 ; au06 ) .
lense thirring precession of this _ same _ hot flow can produce the qpo , with frequency set by the _
same _ parameters of @xmath5 and @xmath38 , together with the surface density of the flow ( idf09 ) .
the surface density is itself given self consistently by mass conservation from the propagating fluctuations .
this predicts that the surface density fluctuates , so predicts that the qpo frequency will vary on short timescales ( i.e. that it is a quasi rather than true period ) .
these fluctuations set an upper limit to the coherence of the qpo , and this increases ( i.e. width decreases ) as @xmath5 decreases .
this is due to the decrease in fluctuation power due to the smaller range of radii from which to pick up variability .
all these features are well known properties of the data ( e.g. remillard & mcclintock 2006 ; dgk07 ) : this model gives the first quantitative description of their origin .
the fluctuations also predict that the flux and qpo frequency are correlated on short timescales , as a perturbation in the surface density at large radii leads to a longer qpo frequency but has little effect on the luminosity .
as this propagates down , it weights the mass distribution to smaller radii , increasing the qpo frequency but also increasingly contributing to the luminosity due to the centrally peaked emissivity .
this behavior has also recently been observed ( heil et al 2011 ) .
the model also gives a framework in which to interpret some otherwise very puzzling aspects of the energy dependence of the variability seen in bhb .
the extended emission region can be inhomogeneous , with different parts of the flow producing a different spectrum .
the outermost parts of the flow are closest to the cool disc , so will intercept more seed photons and have a softer spectrum than that produced in the more photon - starved inner part of the flow ( kawabata & mineshige 2010 ; makishima et al 2008 ; takahashi et al 2008 ) .
this implies that a larger fraction of the lower energy compton scattered photons come from larger radii in the flow than the higher energy ones .
the higher frequency variability is preferentially produced at the smallest radii , where the spectrum is hardest .
the flow at these small radii is also furthest from the cool disc , so has little reflection spectrum superimposed on the compton continuum .
thus the model predicts that the fastest variability has the hardest spectrum and smallest reflected fraction , while slower variability has a softer spectrum and larger reflected fraction .
this trend is also observed in the data ( revnivtsev et al 1999 ) , and is very difficult to interpret in any other geometric picture as the inner disc edge can not change in radius on even the longest timescale ( few seconds ) over which this relation is seen . similarly , the propagating fluctuations model means that a fluctuation starts at larger radii and then accretes down to smaller radii .
thus the fluctuation first affects the region producing a softer spectrum , then propagates down to smaller radii which produce the harder spectrum , so the hard band lags the soft band .
the size of this lag depends on the frequency of fluctuations considered .
slow fluctuations ( low frequencies ) are produced at the outermost radius , so have the longest propagation time down to the innermost radius .
high frequencies are produced only close to the inner radius , so only have a short distance to travel and hence have shorter lags .
this gives rise to the frequency dependent time lags seen in the spectrum ( miyamoto & kitamoto 1989 ; revnivtsev et al 2001 ; k01 ; au06 ) . in future work
, we will quantitatively test these ideas against real data by incorporating a radial dependence of the emitted spectrum in our model , as well as putting in the more complex form for the radial dependence of the viscous frequency predicted from global adaf models and use improved fit statistics ( ingram & done , in preparation ) .
we will also calculate the light curve shape predicted by lense - thirring precession of a hot inner flow , where the comptonised emission is modulated by the difference in projected area of the ( translucent ) hot flow , self occultation and variation in seed photons from the different projected area of the disc ( ingram , done & zycki in prep ) .
this predicts non - sinusoidal variability in the light curve i.e. this predicts the harmonic structure of the qpo . while the many successes of the model are clearly evident , it is also clear that it is still far from complete .
the most obvious outstanding issues are of the interaction of the hot flow with the truncated disc .
the mechanism by which the cool disc truncates is not well established , though evaporation powered by thermal conduction between the two different temperature fluids almost certainly plays some role in this ( liu et al 1997 ; rozanska & czerny 2000 ; mayer & pringle 2007 ) .
whatever the mechanism , it seems physically unlikely that this will give a smooth transition between a cool thin disc and the hot flow .
any inhomogeneities will probably also be amplified by the difference in velocity between the disc and flow ( discs are close to keplarian , while the hot flow is strongly sub keplarian ) so there will be a shearing turbulent layer formed between them .
recent results show that there is variability associated with the truncated disc at a few 10s of seconds in the low / hard state of the bright bhbs gx339 - 4 and swift j1753.5 - 0127 ( wilkinson & uttley 2009 ) , suggesting that there is considerable complexity in the disc truncation ( see also chiang et al 2010 ) .
full numerical simulations of the mri in a composite truncated disc / hot inner flow geometry are probably required in order to show the effect of these
. however , such simulations are way beyond current computer capabilities .
a more tractable issue is the effect of relativity on the propagating fluctuations .
near the black hole , light bending and time dilation should be important and consequently future versions of this model need to take these effects into account .
the final goal should of course be the creation of a fully relativistic model which can produce a fourier resolved spectrum with both energy and time dependence such that we can test it against observations such as the psd , the energy spectrum , the lag spectrum , the cross spectrum etc .
this is of course very ambitious but it is the only way we can genuinely achieve a full theoretical understanding of what drives mass accretion and emission in bhbs .
ai acknowledges the support of an stfc studentship . 99 agol e. , krolik j. h. , 2000 , apj , 528 , 161 arvalo p. , uttley p. , 2006 ,
mnras , 367 , 801 balbus s. a. , hawley j. f. , 1998 , rvmp , 70 , 1 beckwith k. , hawley j. f. , krolik j. h. , 2008 , mnras , 390 , 21 belloni t. , psaltis d. , van der klis m. , 2002 , apj , 572 , 392 cabanac r. a. , valls - gabaud d. , lidman c. , 2008 , mnras , 386 , 2065 chiang c. y. , done c. , still m. , godet o. , 2010 , mnras , 403 , 1102 churazov e. , gilfanov m. , revnivtsev m. , 2001 , mnras , 321 , 759 dexter j. , fragile p. c. , 2011 , apj , 730 , 36 done c. , gierliski m. , kubota a. , 2007 , a&arv , 15 , 1 fragile p. c. , mathews g. j. , wilson j. r. , 2001 , apj , 553 , 955 fragile p. c. , blaes o. m. , anninos p. , salmonson j. d. , 2007 , apj , 668 , 417 fragile p. c. , 2009 , apj , 706 , l246 fragile p. c. , meier d. l. , 2009 , apj , 693 , 771 frank j. , king a. , raine d. , 1992 , apa .. book , gammie c. f. , popham r. , 1998 , apj , 498 , 313 gierliski m. , done c. , page k. , 2008 , mnras , 388 , 753 gierliski m. , nikoajuk m. , czerny b. , 2008 , mnras , 383 , 741 gilfanov m. , revnivtsev m. , molkov s. , 2003 , a&a , 410 , 217 gilfanov , m. , & arefiev , v. 2005 , arxiv : astro - ph/0501215 heil l. m. , vaughan s. , uttley p. , 2010 ,
arxiv , arxiv:1011.6321 henisey k. b. , blaes o. m. , fragile p. c. , ferreira b. t. , 2009 , apj , 706 , 705 ingram a. , done c. , fragile p. c. , 2009 , mnras , 397 , l101 ingram a. , done c. , 2010 , mnras , 405 , 2447 kawabata r. , mineshige s. , 2010 , pasj , 62 , 621 klein - wolt m. , van der klis m. , 2008 , apj , 675 , 1407 kotov o. , churazov e. , gilfanov m. , 2001 , mnras , 327 , 799 krolik j. h. , hawley j. f. , 2002 , apj , 573 , 754 lachowicz p. , done c. , 2010 , a&a , 515 , a65 liu s. , melia f. , 2002 , apj , 573 , l23 lyubarskii y. e. , 1997 , mnras , 292 , 679 liu b. f. , meyer f. , meyer - hofmeister e. , 1997 , a&a , 328 , 247 makishima k. , et al . , 2008 , pasj , 60 , 585 markowitz a. , et al . , 2003 , apj , 593 , 96 mayer m. , pringle j. e. , 2007 , mnras , 376 , 435 merloni a. , vietri m. , stella l. , bini d. , 1999 , mnras , 304 , 155 misra r. , zdziarski a. a. , 2008 , mnras , 387 , 915 miyamoto s. , kitamoto s. , 1989 , natur , 342 , 773 mueller , m. , madejski , g. , done , c. , & zycki , p. 2004
, x - ray timing 2003 : rossi and beyond , 714 , 190 mueller m. , madejski g. , 2009 , apj , 700 , 243 narayan r. , yi i. , 1995 , apj , 452 , 710 narayan r. , kato s. , honma f. , 1997 , apj , 476 , 49 negoro h. , matsuoka m. , mihara t. , otani c. , wang t. g. , awaki h. , 2000 , adspr , 25 , 481 noble s. c. , krolik j. h. , 2009 , apj , 703 , 964 papadakis i. e. , lawrence a. , 1993 , mnras , 261 , 612 psaltis d. , norman c. , 2000 , astro , arxiv : astro - ph/0001391 rao f. , belloni t. , stella l. , zhang s. n. , li t. , 2010 , apj , 714 , 1065 remillard r. a. , mcclintock j. e. , 2006 , aas , 38 , 903 remillard r. a. , sobczak g. j. , muno m. p. , mcclintock j. e. , 2002 , apj , 564 , 962 revnivtsev m. , gilfanov m. , churazov e. , 1999 , a&a , 347 , l23 revnivtsev m. , gilfanov m. , churazov e. , 2001 , a&a , 380 , 520 reynolds c. s. , miller m. c. , 2009 , apj , 692 , 869 raska a. , czerny b. , 2000 , mnras , 316 , 473 schnittman j. d. , 2005 , apj , 621 , 940 schnittman j. d. , homan j. , miller j. m. , 2006 , apj , 642 , 420 shakura n. i. , sunyaev r. a. , 1973 , a&a , 24 , 337 sobczak g. j. , mcclintock j. e. , remillard r. a. , cui w. , levine a. m. , morgan e. h. , orosz j. a. , bailyn c. d. , 2000 , apj , 544 , 993 sobolewska m. a. , ycki p. t. , 2006 , mnras , 370 , 405 stella l. , vietri m. , 1998 , apj , 492 , l59 takahashi h. , et al . , 2008 , pasj , 60 , 69 timmer j. , koenig m. , 1995 , a&a , 300 , 707 titarchuk l. , osherovich v. , 1999 , apj , 518 , l95 uttley p. , mchardy i. m. , 2001 , mnras , 323 , l26 uttley p. , mchardy i. m. , papadakis i. e. , 2002 , mnras , 332 , 231 van der klis m. , 1989 , ara&a , 27 , 517 van der klis m. , 2004 , adspr , 34 , 2646 wagoner r. v. , silbergleit a. s. , ortega - rodrguez m. , 2001 , apj , 559 , l25 wijnands r. , van der klis m. , 1999 , apj , 514 , 939 wilkinson t. , uttley p. , 2009
, mnras , 397 , 666 wilson c. d. , done c. , 2001 , mnras , 325 , 167
here , we attempt to address some of the statistical problems encountered in section [ sec : data ] by focusing in on observation 4 as it is a particularly striking example .
the fit for the stressed version of the model ( hereafter case 1 ) is plotted in the 4th panel from the top in figure [ fig : r - gammas ] and the fit for the stress free version ( hereafter case 2 ) is plotted in the 4th panel from top in figure [ fig : r - gammasf ] . by eye ,
these fits both look excellent and are very difficult to tell apart .
however , the rejection probability is @xmath174 for case 1 and @xmath185 for case 2 ! this is extremely confusing , especially considering that case 2 actually has a slightly lower @xmath155 ( @xmath191 compared to @xmath192 ) than case 1
! we will also discuss the obvious over fitting implied by such low @xmath155 values later on in this section . in an attempt to understand this discrepancy
, we look a little closer at these two examples .
it turns out that @xmath164 for case 2 is larger than in case 1 ( @xmath195 for case 2 ; @xmath196 for case 1 ) .
although this sheds a little light on the mystery , it does nt explain such a huge descrepancy in rejection propbability .
however , the @xmath163 values are all much smaller for case 2 meaning that the simulated values agree with each other much better than they do with the observed values , making it extremely unlikely that the observed points lie on the distribution of simulated points . in an attempt to see this difference
, we plot all of the raw periodogram points for both cases in figure [ fig : scatter ] .
the black and red points are observed and simulated respectively with case 1 on the top and case 2 on the bottom .
we see that the two plots are indistinguishable . in figure
[ fig : dis2 ] , we have looked at 3 specific frequencies : @xmath197hz ( black ) , @xmath198hz ( green ) and @xmath199hz ( blue ) and created a histogram to see the distribution of periodogram values .
note , the three distributions have been normalised to have a mean of @xmath147 and a standard deviation of @xmath200 for clarity . again
, case 1 is on the top .
although there are departures from gaussian behaviour in all of these distributions , it is clear that the distributions for case 2 are far more centrally peaked than those of case 1 .
this is consistent with @xmath163 being smaller in general for case 2 .
there are also hints of baises which may account for the apparent over fitting implied by such small @xmath155 values . in our analysis , we calculated the logarithmically smoothed periodogram over @xmath151 realisations in order to calculate and minimise @xmath155 .
we did this because papadakis & lawrence ( 1993 ) show that the logarithmically smoothed periodogram can have a gaussian distribution if it is smoothed over more than @xmath154 realisations , in comparision to the smoothed periodogram ( i.e. no logarithms ) where you would need more than @xmath201 realisations .
this implies that our logarithmically smoothed periodograms have a gaussian error distribution , therefore making @xmath155 an appropriate fit statistic . however , there are two problems with this , both arising because the error distribution depends on the underlying power spectrum .
papadakis & lawrence ( 1993 ) were looking at red noise which is a significantly different variability process than that considered here .
therefore , just because 20 realisations are enough for them , does nt mean it is enough for us .
more significantly , the same authors show that the logarithmically smoothed periodogram may give a baised estimate of a power spectrum with a high curvature .
since , we are simulating qpos , we are in this very regeim which could explain the hint of bais displayed in figure [ fig : dis2 ] . for these reasons , in a future paper ( ingram & done in prep )
we will use smoothed periodograms ( no logarithms ) and attempt to find a way around the resulting non - gaussian errors . to do this
, we will consider statistical methods more general than @xmath155 but less computationally intensive than minimising the rejection probability .
hopefully this will allows us to overcome these statistical problems . | the power spectra of black hole binaries have been well studied for decades , giving a very detailed phenomenological picture of the variability properties and their correlation with the energy spectrum ( spectral state ) of the source . here
we take the truncated disc / hot inner flow picture which can describe the spectral changes , and show that propagating mass accretion rate fluctuations in the hot flow can match the broad band power spectral properties seen in black hole binaries , i.e. give approximately band limited noise between a low and high frequency break .
the low frequency break marks the viscous timescale at the outer edge of the hot inner flow , which is the inner edge of the truncated disc .
the fluctuations in mass accretion rate propagate towards the central object in a finite time meaning the high frequency break is more complex than simply the viscous timescale at the inner edge of the hot flow because fluctuations on timescales shorter than the propagation time are incoherent .
the model also predicts the lense - thirring precession timescale of the hot flow , as this is set by the combination of inner and outer radius of the flow , together with its surface density which is self consistently calculated from the propagating fluctuations .
we show that this naturally gives the observed relation between the low frequency break and qpo frequency as the outer radius of the flow moves inwards , and that this model predicts many of the observed qpo properties such as correlation of coherence with frequency , and of the recently discovered correlation of frequency with flux on short timescales .
we fit this total model of the variability to a sequence of 5 observed power spectra from the bright black hole binary xte j1550 - 564 as the source transitioned from a low / hard to very high state .
this is the first time that a power spectrum from a black hole binary has been fit with a physical model for the variability .
the data are well fit if the inner radius of the flow remains constant , while the outer radius sweeps inwards from @xmath0 .
this range of radii is the same range as required by models of the energy spectral evolution , giving the first self consistent description of the evolution of both the spectrum and variability of bhb .
= -0.5 cm [ firstpage ] x - rays : binaries accretion , accretion discs |
let @xmath0 $ ] be the ring of polynomials in @xmath1 variables with real coefficients .
a * preordering * of a general ring @xmath2 ( commutative with @xmath3 ) is a subsemiring of @xmath2 which contains the squares . in other words ,
a preordering of @xmath2 is a subset of @xmath2 which contains all @xmath4 , @xmath5 , and is closed under addition and multiplication . for a finite subset @xmath6 of @xmath7 , we write @xmath8 for the preordering of @xmath7 generated by @xmath9 , and @xmath10 for the set of all @xmath11 satisfying @xmath12 ( the basic closed semialgebraic set defined by @xmath9 ) .
note that @xmath10 is uniquely determined by @xmath8 , but typically @xmath8 is not uniquely determined by @xmath10 . for a subset @xmath13 of @xmath14 , we write @xmath15 for the set of all elements of @xmath7 that are nonnegative on @xmath13 .
we always have that @xmath16 .
the preordering @xmath8 is said to be * saturated * if @xmath17 . in this paper
we investigate what geometric properties of @xmath9 imply that @xmath8 is saturated .
this line of investigation has been pursued by scheiderer in a series of papers . in @xcite
, scheiderer showed that @xmath8 is never saturated if @xmath18 .
the case @xmath19 is fairly well understood ; see @xcite , @xcite , @xcite , @xcite .
we focus here on the 2-dimensional case , more precisely , on the affine 2-dimensional case , i.e. , @xmath20 .
we consider only the compact case . in the non - compact case little
is known ; see ( * ? ? ?
* open problem 6 ) and ( * ? ?
* remark 3.16 ) . by
* remark 6.7 ) , @xmath8 is not saturated if @xmath10 contains a two - dimensional cone . in the compact case
, we have the following result of scheiderer ( * ? ? ?
3.3 ) : [ sat3 ] let @xmath6 be irreducible polynomials in @xmath21 $ ] , let @xmath22 be the plane affine curve @xmath23 ( @xmath24 )
. assume : 1 .
@xmath10 is compact 2 .
@xmath22 has no real singular points ( @xmath24 ) 3 .
the real points of intersection of any two of the @xmath22 are transversal , and no three of the @xmath22 intersect in a real point
. then @xmath8 is saturated .
the main goal of this paper is to show that saturation holds in certain other compact cases as well , e.g. , if @xmath25 or @xmath26 . in these examples , the boundary curves @xmath27 and @xmath28 share a common tangent at the origin , so theorem [ sat3 ] does not apply
the fact that saturation holds in these examples is a consequence of our main result , corollary [ c7 m ] , which is an extension of theorem [ sat3 ] .
our original motivation comes from examples which arise naturally while studying semialgebraic sets @xmath29 described by a set @xmath30 of polynomials invariant under an action of a finite group @xmath31 .
the corresponding preordering @xmath32 will typically not be saturated but it can still be `` saturated for invariant polynomials '' ( we refer to this as @xmath31-saturation " ) .
the orbit map @xmath33 ( see @xcite ) relates the @xmath31-saturation of @xmath32 to the saturation of certain preordering @xmath34 corresponding to @xmath35 . in many cases ,
the latter follows from our corollary [ c7 m ] .
an example is given in section 3 .
at the same time , corollary [ c7 m ] does not cover all interesting cases ; in the concluding remarks , we consider some of the remaining cases .
we focus on the case of a compact basic closed semialgebraic set . in ( * ? ? ?
* cor . 3.17 )
, scheiderer proves a useful ` local - global ' criterion , extending ( * ? ? ?
3 ) , for deciding when a polynomial non - negative on a compact basic closed semialgebraic set lies in the associated preordering of the polynomial ring : [ t1 m ] suppose @xmath36 , the subset @xmath13 of @xmath37 defined by the inequalities @xmath38 , @xmath39 is compact , @xmath40 on @xmath13 , and @xmath41 has just finitely many zeros in @xmath13 . then the following are equivalent : 1 .
@xmath41 lies in the preordering of @xmath7 generated by @xmath42 .
2 . for each zero @xmath43 of @xmath41 in @xmath13
, @xmath41 lies in the preordering of the completion of @xmath7 at @xmath43 generated by @xmath42 . in the two dimensional case
this allows one to show that certain finitely generated preorderings are saturated ; see @xcite .
for example , theorem [ sat3 ] can be obtained by combining theorem [ t1 m ] with the following result for power series rings , using the transfer principle : [ t2 m ] suppose @xmath44 $ ] . 1 . if @xmath45 at each ordering of @xmath46 then @xmath41 is a sum of squares in @xmath47 $ ] .
2 . if @xmath45 at each ordering of @xmath46 satisfying @xmath48 then @xmath41 lies in the preordering of @xmath47 $ ] generated by @xmath49 .
3 . if @xmath50 at each ordering of @xmath46 satisfying @xmath51 and @xmath52 then @xmath41 lies in the preordering of @xmath47 $ ] generated by @xmath49 and @xmath53 .
\(1 ) is well - known .
it can be proved using a modification of the analytic argument given in ( * ? ? ?
the proof shows , in fact , that @xmath41 is a sum of two squares . see ( * ? ? ? * th .
1.6.3 ) for more details . ( 2 )
( resp . , ( 3 ) ) follows immediately from ( 1 ) by going to the extension ring @xmath54 $ ] ( resp .
, to the extension ring @xmath55 $ ] ) .
e.g. , to prove ( 2 ) , apply ( 1 ) to @xmath54 $ ] to deduce @xmath56 , @xmath57 $ ]
. decomposing @xmath58 , @xmath59 $ ] , and expanding , yields @xmath60 .
we will prove the following extension of theorem [ t2 m ] .
[ t3 m ] suppose @xmath61 $ ] and @xmath1 is a positive integer . 1 .
if @xmath50 at each ordering of @xmath46 satisfying @xmath62 and @xmath63 then @xmath41 lies in the preordering of @xmath64 $ ] generated by @xmath53 and @xmath65 .
2 . if @xmath50 at each ordering of @xmath66 satisfying @xmath51 , @xmath52 and @xmath67 then @xmath41 lies in the preordering of @xmath64 $ ] generated by @xmath49 , @xmath53 and @xmath68 .
[ rm5 ] suppose @xmath1 is odd , @xmath69 .
then : 1 . for every ordering of @xmath47 $ ] , @xmath70 and @xmath71 @xmath72 @xmath73 , but @xmath49 is not in the preordering of @xmath47 $ ] generated by @xmath53 and @xmath68 .
this shows that an obvious attempt to strengthen theorem [ t3 m ] fails .
2 . similarly ,
for every ordering of @xmath47 $ ] , @xmath74 @xmath72 @xmath75 , but @xmath49 is not in the preordering of @xmath47 $ ] generated by @xmath76 .
note : going to the extension ring @xmath77 $ ] , we see that assertions ( i ) and ( ii ) are essentially equivalent .
we postpone the proof of theorem [ t3 m ] to section [ proof ] .
for now we only explain how theorems [ t1 m ] , [ t2 m ] and [ t3 m ] can be combined to yield the promised extension of theorem [ sat3 ] : [ c7 m ] let @xmath78 be irreducible polynomials in @xmath21 $ ] .
suppose that @xmath79 is compact , and , for each boundary point @xmath43 of @xmath13 , either 1 .
there exists @xmath80 such that @xmath43 is a non - singular zero of @xmath81 , and @xmath13 is defined locally at @xmath43 by the single inequality @xmath38 ; or 2 .
there exists @xmath82 such that @xmath43 is a non - singular zero of @xmath81 and @xmath83 , @xmath81 and @xmath83 meet transversally at @xmath43 , and @xmath13 is defined locally at @xmath43 by @xmath84 , @xmath85 ; or 3 .
there exists @xmath82 such that @xmath43 is a non - singular zero of @xmath81 and @xmath83 , @xmath81 and @xmath83 share a common tangent at @xmath43 but do not cross each other at @xmath43 , and @xmath13 is described locally at @xmath43 as the region between @xmath86 and @xmath87 ; or 4 .
there exists @xmath88 such that @xmath43 is a non - singular zero of @xmath81 , @xmath83 and @xmath89 , @xmath81 and @xmath83 share a common tangent at @xmath43 , @xmath81 and @xmath89 meet transversally at @xmath43 , and @xmath13 is described locally at @xmath43 as the part of the region between @xmath23 and @xmath87 defined by @xmath90 .
then the preordering of @xmath21 $ ] generated by @xmath42 is saturated .
let @xmath91 denote the preordering of @xmath21 $ ] generated by @xmath42 .
we wish to show that @xmath92 $ ] , @xmath45 on @xmath13 @xmath72 @xmath93 .
we may assume @xmath94 , @xmath95 .
the hypothesis implies , in particular , that @xmath13 is the closure of its interior .
this allows us to reduce further to the case where @xmath41 is square - free and @xmath96 for each @xmath80 .
in this situation , @xmath41 has only finitely many zeros in @xmath13 , so theorem [ t1 m ] applies , i.e. , to show @xmath93 , it suffices to show that , for each zero @xmath43 of @xmath41 in @xmath13 , @xmath41 lies in the preordering of the completion of @xmath21 $ ] at @xmath43 generated by @xmath42 .
if @xmath43 is an interior point of @xmath13 this follows from theorem [ t2m](1 ) .
if @xmath43 is a boundary point of @xmath13 satisfying ( 1 ) ( resp . ,
( 2 ) , resp . ,
( 3 ) , resp . , ( 4 ) ) then it follows from theorem [ t2m](2 ) ( resp . , theorem [ t2m](3 ) , resp . ,
theorem [ t3m](1 ) , resp .
, theorem [ t3m](2 ) ) .
we use the transfer principle and apply theorems [ t2 m ] and [ t3 m ] with @xmath97 , @xmath98 , where @xmath99 are suitably chosen local parameters at @xmath43 .
if @xmath43 is an interior point of @xmath13 we choose @xmath100 , @xmath101 where @xmath102 . in case
( 1 ) , we choose local parameters @xmath103 , @xmath104 with @xmath105 . in case ( 2 ) , we choose local parameters @xmath103 , @xmath104 with @xmath105 , @xmath106 . in case ( 3 ) , choose local parameters @xmath103 , @xmath81 . by the preparation theorem (
* cor . 1 ,
p. 145 ) , @xmath107 for some unit @xmath108 , some @xmath109 and some unit @xmath110 $ ] .
then @xmath111 where @xmath112 and @xmath113 . by the geometry of the situation , the units @xmath114
are positive units and @xmath1 is even .
take @xmath115 , so @xmath116 , and apply theorem [ t3m](1 ) . in case
( 4 ) choose local parameters @xmath103 , @xmath81 with @xmath117 .
as before , this yields @xmath111 for some units @xmath114 and some @xmath118 . by the geometry of the situation , @xmath114 are positive units .
take @xmath115 , so @xmath116 , and apply theorem [ t3m](2 ) .
if @xmath119 and @xmath120 then @xmath121 is the unit square . note that @xmath8 is saturated , by theorem [ sat3 ] . on the other hand
, it can be easily verified that @xmath122 , hence @xmath32 is not saturated .
let @xmath123 be the fourth dihedral group acting on @xmath124 and @xmath21 $ ] in a standard way " .
for every @xmath31-invariant subset @xmath125 of @xmath21 $ ] write @xmath126 .
we would like to show that @xmath32 is * @xmath31-saturated * , i.e. @xmath127 or equivalently , @xmath128 .
clearly , @xmath21^g$ ] is an @xmath129-algebra containing @xmath130 it can be shown that @xmath131 and @xmath132 are algebraically independent and that they generate @xmath21^g$ ] .
hence , the mapping @xmath133
\to \r[x , y]^g , \quad \tilde{\pi}(f)(u , v ) = f(u(x , y),v(x , y))\ ] ] is an isomorphism . on the other hand ,
the mapping @xmath134 is not onto .
it is easy to see that @xmath135 the mapping @xmath33 is not one - to - one either .
it can be shown that two points have the same image if and only if they lie in the same @xmath31-orbit .
( we call @xmath33 the * orbit map * and @xmath136 the * orbit space*. ) the set @xmath137 ( picture on the left ) contains exactly one point from each orbit of @xmath10 .
now we can compute @xmath138 ( picture on the right ) by either parametrizing the boundary of @xmath139 or the following way : @xmath140 by corollary [ c7 m ] , the preordering @xmath141 is saturated . hence @xmath142
assertion ( 2 ) follows from assertion ( 1 ) , by going to the extension ring @xmath54 $ ] , so it suffices to prove ( 1 ) .
we can assume @xmath143 .
we know @xmath47 $ ] is a ufd ( * ? ? ?
6 , p. 148 ) .
factor @xmath41 into irreducibles in @xmath47 $ ] . using the preparation theorem
, we can assume the factorization has the form @xmath144 where @xmath145 is a unit and each @xmath146 is a monic polynomial in @xmath53 with coefficients in @xmath147 $ ] , with all coefficients except the leading coefficient in the maximal ideal of @xmath147 $ ] .
we can reduce to the case where @xmath148 or @xmath3 and @xmath149 has no repeated irreducible factors . since @xmath150 is a square in @xmath47 $ ] , we can assume further that @xmath151 .
since @xmath53 and @xmath65 are obviously in the preordering generated by @xmath53 and @xmath65 , we can assume @xmath152 and @xmath153 .
more generally , if @xmath149 has an irreducible factor @xmath43 which has constant sign on the set @xmath154 in the real spectrum ( see @xcite ) of @xmath46 then , by part ( 2 ) of theorem [ t2 m ] , @xmath155 is in the preordering generated by @xmath53 .
similarly , if @xmath43 has constant sign on the set @xmath156 in the real spectrum of @xmath46 then , by part ( 2 ) of theorem [ t2 m ] ( using the fact that @xmath47 = \r[[x , x^{2n}-y]]$ ] ) , @xmath155 is in the preordering generated by @xmath65 .
consequently , we can assume that @xmath149 has no such irreducible factors .
fix an irreducible factor @xmath43 of @xmath149 and consider the discrete valuation on @xmath46 with associated valuation ring @xmath47_{(p)}$ ] .
the residue field is @xmath157}{(p ) } = \frac{\r((x))[y]}{(p)}$ ] ( * ? ? ?
6 , p. 148 ) .
set @xmath158 . since @xmath159 , @xmath160 , we know that @xmath161 , @xmath162 .
@xmath163 is a finite extension of the complete discrete valued field @xmath164 so it either has no orderings ( if the residue field is @xmath165 ) or two orderings ( if the residue field is @xmath129 ) .
claim 1 : @xmath163 has no ordering satisfying @xmath166 . otherwise , pulling this ordering back to @xmath46 , using baer - krull , yields two orderings on @xmath46 satisfying @xmath167 , one with @xmath168 and one with @xmath169 . since
an irreducible factor @xmath170 of @xmath41 different from @xmath43 has the same sign at each of these two orderings , and since @xmath43 has multiplicity 1 in @xmath41 , one of these two orderings must make @xmath171 .
this contradicts our assumption and proves the claim .
claim 2 : @xmath163 has an ordering satisfying @xmath172 and also an ordering satisfying @xmath173 . by assumption @xmath174 is not always positive on the set @xmath154 in the real spectrum of @xmath46 , so there exists an ordering of @xmath46 , with real closure @xmath175 say , with @xmath154 and @xmath176 , so the polynomial @xmath177 ( obtained by replacing @xmath53 by the new variable @xmath178 ) has a root @xmath179 in @xmath175 .
then @xmath180 defines an @xmath164-embedding of @xmath163 into @xmath175 , so @xmath163 has an ordering satisfying @xmath181 , i.e. , @xmath172 .
we prove the second assertion when @xmath182 is odd .
the proof when @xmath182 is even is similar . by assumption @xmath43
is not always negative on the set @xmath156 in the real spectrum of @xmath46 , so there exists an ordering of @xmath46 with real closure @xmath175 say , with @xmath183 and @xmath184 , so the polynomial @xmath177 has a root @xmath185 in @xmath175 .
then @xmath180 defines an @xmath164-embedding of @xmath163 into @xmath175 , so @xmath163 has an ordering satisfying @xmath186 , i.e. , @xmath187 .
denote the valuation on @xmath163 by @xmath188 .
since @xmath189 we see that @xmath190 .
since @xmath163 has an ordering satisfying @xmath191 , it follows that @xmath192 . at the same time , @xmath193 is not possible .
( if @xmath194 then @xmath195 , @xmath145 a unit .
since @xmath104 is positive at one ordering and negative at the other , the same would be true for @xmath145 , which is not possible . )
thus @xmath196 . of course
, since the various roots @xmath197 of @xmath43 in the algebraic closure of @xmath164 are conjugate to @xmath104 over @xmath164 , they all have the same value @xmath198 .
write @xmath199 where the @xmath200 are irreducible , @xmath201 , @xmath202 , @xmath203 , @xmath204 , where @xmath205 is a fixed root of @xmath200 .
we know @xmath206 .
decompose @xmath41 as @xmath207 where @xmath208 , @xmath209 and @xmath210 .
claim 3 : @xmath211 is positive at both orderings of @xmath164 , i.e. , @xmath211 is a square in @xmath147 $ ] .
suppose to the contrary that @xmath211 is negative at one of the orderings of @xmath164 .
consider the discrete valuation on @xmath46 with valuation ring @xmath47_{(y)}$ ] and residue field @xmath164 . pulling the culprit ordering of @xmath164 back to @xmath46 , using baer - krull , yields two orderings of @xmath46 ,
one of which satisfies @xmath212 and @xmath213 .
this is a contradiction .
we write each term @xmath214 , @xmath215 in ( [ eq1 ] ) as @xmath216 factoring in the obvious way , we see that @xmath217 lies in the preordering generated by @xmath65 and @xmath53 . to complete the proof
, it suffices to show we can choose the elements @xmath218 $ ] , @xmath215 , so that @xmath219 are squares in @xmath147 $ ] .
since @xmath215 , @xmath220 we choose the @xmath221 as follows : if @xmath222 or @xmath223 and @xmath224 , then @xmath225 has positive value . in this case , we choose @xmath221 with small positive lowest coefficient and with @xmath226 in the remaining case , where @xmath148 and @xmath227 , @xmath228 , @xmath229 , and we choose @xmath230 .
the point is , with this choice of @xmath221 , for each @xmath215 , either @xmath231 has larger value than @xmath211 or , it has the same value as @xmath211 , but its lowest coefficient is small .
\1 . theorems [ t2 m ] and [ t3 m ] do not cover all interesting cases .
the general question remains : when is a finitely generated preordering of @xmath47 $ ] saturated ?
recall that the * saturation * of a preordering @xmath91 of a general ring @xmath2 ( commutative with 1 ) is the intersection of all orderings of @xmath2 containing @xmath91 , and that @xmath91 is said to be * saturated * if it coincides with its saturation .
saturation in case ( i ) is a consequence of saturation in case ( ii ) , by going to the extension ring @xmath77 $ ] . in an analogous way ,
saturation in case ( ii ) is a consequence of ( * ? ? ?
5.1 ) , by going to the extension ring @xmath234[z]}{(z^2-y^2+x^n ) } = \frac{\r[[x , y , z]]}{(z^2-y^2+x^n)}.\ ] ] ( * ? ? ?
5.1 ) asserts that the ring @xmath2 defined above satisfies psd = sos , i.e. , that the preordering of @xmath2 consisting of sums of squares in saturated . actually , ( * ? ? ?
5.1 ) is stated in terms of analytic function germs .
what we are quoting here is the formal power series version of the result .
note : knowing saturation holds in case ( i ) allows one to extend corollary [ c7 m ] , adding an additional case to the list .
3 . it is still not known if the following preorderings of @xmath47 $ ] are saturated : 1 .
the preordering of @xmath47 $ ] generated by @xmath235 and @xmath236 , @xmath1 odd , @xmath237 even , @xmath238 .
the preordering of @xmath47 $ ] generated by @xmath239 and @xmath240 , @xmath1 odd , @xmath237 even , @xmath238 , @xmath241 $ ] , @xmath242 . a positive answer in cases ( iii ) and ( iv ) , coupled with what we already know by theorems [ t2 m ] and [ t3 m ] and case ( i ) above ,
would complete our understanding of saturation for preorderings of @xmath47 $ ] generated by finitely many elements of order @xmath243 .
the proof of this assertion will not be given here .
the * order * of @xmath244 $ ] is defined to be the greatest integer @xmath245 such that @xmath246 , where @xmath247 denotes the maximal ideal of @xmath47 $ ] .
the case where some of the generators have order @xmath248 seems to be pretty much wide open .
case ( ii ) is of this type , as is the example given earlier , in remark [ rm5 ] ( ii ) .
if @xmath249 $ ] and psd = sos holds for the ring @xmath250[z]}{(z^2-g)}$ ] , then the preordering of @xmath47 $ ] generated by @xmath149 is saturated . combining this with ( * ? ? ?
3.1 ) yields a variety of examples of this sort where @xmath149 has order 2 or 3 and saturation holds .
99 j. bochnak , m. coste , m .- f .
roy , _ gomtrie algbrique relle_. springer verlag 1986
. j. bochnak , j .- j .
risler , le thorme des zros pour les varits analytiques relles de dimension 2 .
sci . cole norm .
_ * 8 * , 353364 ( 1975 ) .
j. cimpri , s. kuhlmann , c. scheiderer , @xmath31-invariant sums of squares and @xmath31-invariant moment problems . to appear in _ trans .
fernando , positive semidefinite germs in real analytic surfaces . _
ann . _ * 322 * , 49 - 67 ( 2002 ) .
s. kuhlmann , m. marshall , positivity , sums of squares and the multi
dimensional moment problem . _ trans .
_ * 354 * , 4285 - 4301 ( 2002 ) .
s. kuhlmann , m. marshall , n. schwartz , positivity , sums of squares and the multi dimensional moment problem ii .
geom . _ * 5 * , 583 - 607 ( 2005 ) .
m. marshall , _ positive polynomials and sums of squares _ , volume 146 of _ mathematical surveys and monographs _ , amer .
d. plaumann , _ bounded polynomials , sums of squares , and the moment problem_. phd thesis , konstanz , 2008 . c. scheiderer , sums of squares of regular functions on real algebraic varieties . _
* 352 * , 10301069 ( 1999 ) . c. scheiderer , sums of squares on real algebraic curves .
z. _ * 245 * , 725760 ( 2003 ) . c. scheiderer , sums of squares on real algebraic surfaces .
_ manuscripta math .
_ * 119 * , 395410 ( 2006 ) .
k. schmdgen , the @xmath13-moment problem for compact semi - algebraic sets .
ann . _ * 289 * , 203206 ( 1991 ) .
o. zariski , p. samuel , _ commutative algebra , vol .
volume 29 of _ graduate texts in math .
_ , springer - verlag 1980 . | we extend and generalize results of scheiderer ( 2006 ) on the representation of polynomials nonnegative on two - dimensional basic closed semialgebraic sets .
our extension covers some situations where the defining polynomials do not satisfy the transversality condition .
such situations arise naturally when one considers semialgebraic sets invariant under finite group actions . |
low - energy anti - neutrinos from nuclear reactors are well suited to determine the neutrino mixing angle @xmath2 , which is important for the search of cp violation in the leptonic sector ; see , e.g. , refs .
the double - chooz @xcite , daya bay @xcite , and reno collaborations @xcite are aiming to measure @xmath2 with very high precision with the use of @xmath3 s produced in nuclear reactors . the present upper bound to this quantity reported by the chooz @xcite and minos @xcite collaborations is : @xmath4 .
the double - chooz , daya bay and reno experiments monitor the inverse beta - decay reaction on a hydrogen target |_e + p e^+ + n [ dcdb ] [ eq : reaction ] for a known anti - neutrino energy flux .
the positron yield is measured as a function of the positron energy .
an accurate extraction of the mixing angle @xmath2 from an analysis of the measured positron yield requires a precise knowledge of the radiative corrections ( rcs ) . in earlier papers @xcite
, the relevant rcs were evaluated in the theoretical framework developed by sirlin and marciano @xcite . in this framework , to be referred to as the s - m approach , the rcs of order @xmath1 are decomposed into so - called outer and inner corrections .
the outer correction is a universal function of the lepton energy and is independent of the details of hadron physics , whereas the inner correction is influenced by short - distance physics and the hadron structure .
the inner corrections coming from @xmath5 and weak - boson loop diagrams are divided into high - momentum and low - momentum parts .
the former is evaluated in the current - quark picture , while the latter is computed with the use of the phenomenological electroweak - interaction form factors of the nucleon .
although the estimates of inner corrections in the s - m formalism are considered to be reliable to the level of accuracy quoted in the literature , the possibility that these estimates may involve some degree of model dependence is not totally excluded .
we present here a calculation of the rcs to order @xmath1 based on effective field theory ( eft ) .
we use heavy - baryon chiral perturbation theory ( hb@xmath6pt ) , which is an effective low - energy theory of qcd , see e.g. ref @xcite . in hb@xmath6pt
the short distance hadronic and electroweak processes are subsumed into a well - defined set of _ low - energy constants _ ( lecs ) .
in other words , these lecs systematically parameterize the inner corrections of the s - m approach .
therefore , insofar as there are enough sources of information to determine the values of these lecs , hb@xmath6pt leads to model - independent results with systematic estimates of higher - order corrections .
the use of hb@xmath6pt to calculate electroweak transition amplitudes for the nucleon and few - nucleon systems were pioneered in refs .
@xcite , and subsequently there have been many important developments . in ref .
@xcite , we presented the first ever eft - based calculation of rcs for the neutron @xmath7-decay process , @xmath8 .
because in hb@xmath6pt the nucleons are treated as point - like , it is expected on general grounds that the order-@xmath1 rcs are common between neutron @xmath7-decay and inverse @xmath7-decay .
meanwhile , it should be mentioned that , in the counting scheme adopted here and in ref.@xcite , the order-@xmath1 rcs are of the same order as the @xmath9 nucleon - recoil corrections including the weak magnetism " contributions , and hence a consistent eft calculation should include these recoil corrections simultaneously .
we present here such an eft calculation , taking advantage of the fact that the @xmath9-expansion is a natural part of our counting scheme and thus dictates how to incorporate recoil corrections order by order ( see later in the text ) .
since exactly the same lecs are involved in the eft calculations of inverse @xmath7-decay and neutron @xmath7-decay , we can in principle use the existing neutron @xmath7-decay data to determine those lecs and make a model - independent estimation of rcs for the inverse @xmath7-decay , the @xmath9 recoil corrections are properly taken into account . in this connection , we note that an attempt has been made in the literature @xcite to directly relate the neutron decay rate with the inverse @xmath7-decay cross section , assuming that the order-@xmath1 corrections ( rcs and recoil corrections combined ) are common between these processes .
as mentioned , this assumption is justified as far as the genuine rcs of order @xmath1 is concerned .
however , as described later in the text , our calculation shows differences between the @xmath9 corrections for inverse @xmath7-decay , eq .
( [ eq : reaction ] ) , and those for neutron @xmath7-decay .
we therefore caution against writing the cross section for the reaction in eq .
( [ eq : reaction ] ) in terms of the neutron mean life , @xmath10 , as advocated in refs .
@xcite .
this paper is organized as follows . in section 2
we explain a theoretical framework to be used and present the results for the order-@xmath1 rcs . in section 3
we consider the recoil corrections and compare our results with an earlier work@xcite .
section 4 gives a summary of our calculations and conclusions .
the appendix describes some technical details concerning the hb@xmath6pt treatment of the infrared singularity .
we use here essentially the same theoretical framework as in ref .
@xcite , in which we calculated rcs for neutron @xmath7-decay .
we therefore give only a brief recapitulation of our formalism , relegating details to ref .
@xcite .
our calculation is based on the @xmath11-expansion scheme , where @xmath12 @xmath13 @xmath14 ( @xmath15 ) represents a typical four - momentum transfer for incident low - energy reactor anti - neutrinos ( @xmath16 mev ) , and @xmath17 gev ( @xmath18 mev is the pion - decay constant ) is the chiral scale .
it is to be noted that the expansion parameter in our scheme is very small and that , as explained in more detail below , the lowest order recoil corrections @xmath19 are of the same order as the lowest order radiative corrections ; _ viz . _ , @xmath20 , where @xmath21 .
the _ leading order _ ( lo ) transition matrix element for the inverse @xmath7-decay , eq.([dcdb ] ) , is evaluated ignoring nucleon recoil and radiative corrections . the _ next - to - leading order _ ( nlo )
corrections in our counting scheme are the recoil corrections ( @xmath22 ) and the radiative corrections ( @xmath23 ) .
the recoil corrections , which include the weak magnetism " term , will be specified in eq.([eq : lnnee ] ) below . for the sake of the transparency of presentation , we shall in this paper separate these corrections from the @xmath9 ( kinematic ) corrections to the phase - space .
the effective lagrangian relevant to our calculation includes the relativistic leptonic weak interaction current and the lo and nlo heavy - baryon lagrangian @xmath24 where @xmath25 ( iv\cdot d ) n \ ; , \label{eq : lnn}\\ { \cal l}_{nn\psi\psi } \!&=&\!\!\!\ ! -\!\left(\frac{g_f v_{ud}}{\sqrt{2}}\right ) \ !
\left [ \bar{\psi}_e \gamma_\mu ( 1-\gamma_5)\psi_\nu \right ] \ !
\left\ { \bar{n } \tau^+ \ !
\left [ \left ( 1\!+\!\frac{\alpha}{4\pi } \ev
\right ) v^\mu - 2g_a\
! \left ( 1\!+\!\frac{\alpha}{4\pi } \ea
\right ) s^\mu \right]\ !
n \right . \nonumber \\
\!&+ & \!\!\ !
\left . \!\ ! \frac{1}{2m_n } \bar{n } \tau^+\!\ ! \left [ i(v^\mu v^\nu\!\ ! -\ !
g^{\mu \nu } ) ( \stackrel{\leftarrow}{\partial } \!-\ !
\stackrel{\rightarrow}{\partial } ) _
\nu \!-\!2i\muv\!\left [ s^\mu\ ! , s\!\cdot\ ! ( \stackrel{\leftarrow}{\partial } \!+\!\stackrel{\rightarrow}{\partial } ) \right ] \!-\!2i\ga v^\mu s\!\cdot\ ! ( \stackrel{\leftarrow}{\partial } \!-\!\stackrel{\rightarrow}{\partial } ) \right]\
! n \right\ } .
\nonumber \\ & & \label{eq : lnnee}\end{aligned}\ ] ] @xmath26 in eq.([eq : lqed ] ) is the usual qed lagrangian , where @xmath27 , and @xmath28 is the covariant derivative ; for the gauge parameter @xmath29 , we use here @xmath30 ( feynman gauge ) .
@xmath31 is the heavy - nucleon lagrangian including the photon - nucleon interaction , and @xmath32 is the low - energy lo and nlo current - current weak interaction .
we give in eq.([eq : lnnee ] ) the explicit forms of nlo nucleon - recoil terms dictated by hb@xmath6pt . in the above
, @xmath33 is the axial coupling constant , while @xmath34 is the nucleon velocity vector , and @xmath35 is the nucleon spin ; they satisfy @xmath36 .
we choose here @xmath37 and @xmath38 . in the nlo part of the lagrangian the nucleon isovector magnetic moment @xmath39 .
the low - energy constants ( lecs ) , @xmath40 , @xmath41 , @xmath42 and @xmath43 , are counter - terms which regulate the ultraviolet ( uv ) divergences of the virtual photon - loop diagrams .
these lecs incorporate the short - range radiative physics that is not probed in a low - energy process .
the lecs , @xmath40 and @xmath41 , are related to the wave - function renormalization factors of the positron and proton , respectively .
the lecs , @xmath42 and @xmath43 , are related to the fermi and gamow - teller amplitudes .
the fermi coupling constant , @xmath44 gev@xmath45 , is determined from muon decay , and the ckm matrix element , @xmath46 , is given by the pdg @xcite . for later convenience , we rearrange the lecs in eq.([eq : lnnee ] ) by rewriting the hadronic part in the first line in eq.([eq : lnnee ] ) as @xmath47 \!n + \left(\frac{\alpha}{4\pi}\right)\!\ev \bar{n}\tau^+ \!\left [ v^\mu \!-\!2\tildega s^\mu \right]\ !
n + { \cal o}(\alpha^2 ) \ ; , \nonumber\end{aligned}\ ] ] where we have introduced the redefined axial coupling constant , @xmath48 $ ] . as in the neutron @xmath7-decay case @xcite , to the order of our concern , @xmath49 can always be replaced by @xmath50 .
this also applies to the nlo recoil contributions since the @xmath9 recoil corrections are of the same order as order-@xmath1 corrections in the adopted counting scheme .
the order-@xmath1 radiative corrections to the nucleon magnetic moments are for the same reason of higher order in our scheme and hence neglected in this work . in this paper
we derive a model - independent expression for the lowest order radiative and recoil corrections to the reaction , @xmath51 , where the four - momentum of each particle is indicated in the parentheses .
we shall concentrate on an experimental setup in which none of the particle spins are monitored by the detector .
there is one subtle aspect of the above reaction which deserves some discussion . in experiments
, the final state positron will always be accompanied by ( often undetected ) soft bremsstrahlung photons .
if the bremsstrahlung photon energy , @xmath52 is less than the detector resolution , @xmath53 , the energy recorded by the detector is the sum of the actual outgoing positron energy , @xmath54 , and the bremsstrahlung photon energy , @xmath52 ; i.e. , @xmath55 is what is measured as the `` detected positron '' energy , with the corresponding `` detected positron '' momentum being @xmath56 .
the two processes we evaluate are @xmath57 and @xmath58 . due to the finite detector resolution
the second bremsstrahlung process is not observed ; it only contributes to the rcs of the first process , i.e. , the soft bremsstrahlung photons are an integral part of the detected positron " .
thus , the first process has become , @xmath59 , where the positron momentum @xmath60 has been replaced by @xmath61 in order to indicate that the soft bremsstrahlung process has been incorporated into this effective " reaction . the cross section for this
effective " reaction is given in terms of the effective invariant amplitude @xmath62 : @xmath63 { \rm d}(\cos\theta_e ) \
, , \label{eq : nlo}\end{aligned}\ ] ] where @xmath64 is the velocity of the outgoing `` detected positron '' for a given incident ( anti-)neutrino beam energy , @xmath65 and detector reading , @xmath66 ; @xmath67 , and @xmath68 is the phase - space factor to be discussed later in the text ( see , eq .
( [ eq : ps - recoil ] ) ) .
the two velocity - dependent functions , @xmath69 ( @xmath70 ) , are written up to nlo as @xmath71 here @xmath72 ( see , eqs .
( [ eq : g1 ] ) and ( [ eq : g2 ] ) ) represent the lowest - order radiative corrections , and @xmath73 ( see , eq .
( [ eq : l - recoil ] ) ) , which will be evaluated in the next section , represent the recoil corrections arising from the lagrangian in eq.([eq : lnnee ] ) .
the calculation of the function @xmath72 is described next . for the analysis of the radiative corrections , we explicitly distinguish between the outgoing positron and the bremsstrahlung photon .
there are two distinct categories of radiative corrections , the bremsstrahlung and the virtual photon loop corrections .
the corresponding feynman diagrams are shown in fig .
[ fig : feynmanqed ] .
since @xmath74 and @xmath75 are of the same order in our counting scheme , the invariant matrix element , @xmath76 for bremsstrahlung is evaluated assuming @xmath77 .
the differential cross section for the radiative process , @xmath78 and @xmath79 denote momenta in the static nucleon approximation , i.e. , @xmath80 , @xmath81 , and correspondingly , @xmath82 .
] , is given by @xmath83 where @xmath84 is the maximum energy of the `` detected positron '' in the static nucleon approximation , i.e. , @xmath85 .
the bremsstrahlung matrix element squared with the static neutron is @xmath86 \left [ 1+\left(\frac{1-g_a^2}{1 + 3g_a^2}\right ) \frac{\tilde{{\mbox{\boldmath $ p$}}}_e\cdot{\mbox{\boldmath $ p$}}_\nu}{\tilde{e}_ee_\nu } \right ] \right .
\nonumber \\ & + & \!\!\ ! ( 1\!+\!3g_a^2 ) \big [ \frac{2\tilde{e}_e^2 + \tilde{e}_e \tilde{e}_\gamma+ \tilde{{\mbox{\boldmath $ p$}}}_e\cdot\tilde{{\mbox{\boldmath $ k$}}}+ \tilde{e}_\gamma^2 } { \tilde{e}_e } -
\ \frac { m_e^2 \tilde{e}_\gamma ( \tilde{e}_e+\tilde{e}_\gamma ) } { \tilde{e}_e(\tilde{p}_e\cdot \tilde{k})}\big ] \nonumber \\ & + & \!\ ! ( 1-g_a^2)\big[(\tilde{{\mbox{\boldmath $ p$}}}_e\cdot{\mbox{\boldmath $ p$}}_\nu)\left ( \frac{2\tilde{e}_e+\tilde{e}_\gamma}{\tilde{e}_e e_\nu } - \frac { m_e^2\tilde{e}_\gamma } { \tilde{e}_e e_\nu ( \tilde{p}_e\cdot \tilde{k})}\right ) \big ] \nonumber \\ & + & \left .
\!\!(1-g_a^2)\big [ ( \tilde{{\mbox{\boldmath $ k$}}}\cdot{\mbox{\boldmath $ p$}}_\nu)\left ( \frac{\tilde{e}_e+ \tilde{e}_\gamma } { \tilde{e}_e e_\nu } - \frac { m_e^2 \tilde{e}_\gamma } { \tilde{e}_e e_\nu ( \tilde{p}_e\cdot \tilde{k})}\right)\big ] \right\ } \ ; .
\label{eq : mbrems2}\end{aligned}\ ] ] we note that the above expression for @xmath87 is identical to that for neutron @xmath7-decay derived in ref
. @xcite .
we also remark that eq.([eq : mbrems2 ] ) was derived earlier by fukugita and kubota @xcite , who used the s - m approach @xcite and a finite photon mass in order to regulate the infrared ( ir ) singularity . in the integration over the bremsstrahlung photon energy in eq.([eq : sigmabr ] ) the maximum photon energy occurs when @xmath88 , i.e. , @xmath89 .. ] in this context , the same question again arises as to whether or not the experiment can distinguish between the two final states , @xmath90 and @xmath91 .
if the detector resolution in the experimental setup is such that one can detect photons with an energy @xmath92 in the interval @xmath93 , we should integrate the bremsstrahlung photon energy @xmath94 from 0 to @xmath53 in eq .
( [ eq : sigmabr ] ) .
however , if the experiment is unable to distinguish these two final states , we should integrate from 0 to @xmath95 . in order to compare with earlier works ,
we concentrate here on the latter case .
the integral over the radiative photon spectrum invariably gives rise to an ir singularity .
we use dimensional regularization to deal with the ir singularity ; some details regarding the bremsstrahlung integral are presented in the appendix .
as is well known , the ir singularity appearing in eq.([eq : sigmabr ] ) is cancelled by the contributions from virtual photon - loop diagrams in accordance with bloch and nordsieck @xcite , see also @xcite .
the evaluation of the loop diagrams in dimensional regularization can be found in the literature , see , _
e.g. , _ ref .
it is notable that , apart from the so - called coulomb factor " @xmath96 , which arises in , e.g. neutron @xmath7-decay from a photon - loop diagram , the matrix element given by these virtual photon loops is identical to the one in neutron @xmath7-decay .
the uv - divergencies originating from the photon loop diagrams are regulated by the lecs in the lagrangian .
these lecs are renormalized by the usual effective field theoretical method based on dimensional regularization of the loop integrals , see e.g. , refs .
the finite lecs renormalized at the scale @xmath97 are @xmath98 + \frac{3}{2}\,{\rm ln}\left(\frac{\mu^2}{m_n^2}\right ) \ ; . \nonumber\end{aligned}\ ] ] the lec , @xmath99 , which was introduced by ando _
et al . _
@xcite in the evaluation of the rc for neutron @xmath7-decay , subsumes short distance physics not probed at low energies and depends on the regularization scale @xmath97 .
combining the bremsstrahlung and virtual photon - loops contributions calculated to order @xmath1 , and noting that @xmath100 , we obtain , neglecting @xmath101 contributions , @xmath102 , @xmath103 , @xmath104 , appearing in eq.([eq : gi ] ) .
dropping terms of @xmath105 , we choose to write the results in the following form : @xmath106 \left[1\!+\!\frac{\alpha}{2\pi}\delta_{out}(\beta)\right ] \label{eq : g1}\end{aligned}\ ] ] @xmath107 \left[1+\frac{\alpha}{2\pi}\tilde{\delta}_{out}(\beta)\right ] \ ; , \label{eq : g2}\end{aligned}\ ] ] where the inner " corrections , which are independent of @xmath7 , are encoded in the lec @xmath108 and defined as @xmath109 .
outer " radiative corrections constitute the well - known , model - independent , long - distance qed corrections that do not contain any hadronic effects , and are given by @xmath110 + \left(\frac{3\beta}{4}\ !
+ \!\frac{7}{4\beta } \right ) { \rm ln}\!\left(\frac{1\!+\!\beta}{1\!-\!\beta}\right ) \label{eq : delta - out}\end{aligned}\ ] ] @xmath111 { \rm ln}\!\left(\frac{1\!+\!\beta}{1\!-\!\beta}\right ) \nonumber \\ & & - 4 \left [ \frac{1}{2\beta } \ , { \rm ln}\!\left(\frac{1\!+\!\beta}{1\!-\!\beta}\right ) \
! -\!1\right ] { \rm ln}\!\left[\left(\frac { 1\!+\!\beta } { 2\beta } \right ) \frac{\sqrt{1\!+\!\beta}+\!\sqrt{1\!-\!\beta } } { \sqrt{1\!+\!\beta}\!-\!\sqrt{1\!-\!\beta } } \
\right ] \ ; .
\label{eq : deltatilde - out}\end{aligned}\ ] ] the above expressions for @xmath112 and @xmath113 reproduce the results obtained by fukugita and kubota @xcite .
we also note that @xmath114 , where @xmath115 is the function introduced by sirlin @xcite .
as mentioned , @xmath99 also appears in the expression for the rc for neutron @xmath7-decay . therefore , it is in principle possible to determine @xmath99 using relevant high - precision low - energy data involving baryons . due to lack of useful experimental data , ando _
et al . _
@xcite determined @xmath99 at @xmath116 by comparing their results for neutron @xmath7-decay with those obtained in the s - m approach @xcite .
as mentioned , these corrections have two different origins .
one comes from the lagrangian itself , and the other arises from the expansion of the kinematic factors in the phase - space integral .
below we treat these two types of recoil corrections separately and compare our results with those in ref .
it is to be noted again that , in evaluating the @xmath101 corrections , we can neglect @xmath117 radiative effects , since @xmath118 terms are of higher order in our counting scheme .
one can , therefore , assume that the outgoing positron energy , @xmath119 , and correspondingly , the positron velocity , @xmath120 .
* kinematic ( phase space ) corrections * + the phase space factor , @xmath68 , appearing in eq.([eq : nlo ] ) to the lowest order ( lo ) in the @xmath9 expansion is given by @xmath121 with the neutron regarded as being static , i.e. , @xmath122 .
to nlo , the above expression for @xmath68 needs to be corrected to incorporate the kinetic energy of the recoil neutron from the relation @xmath123 .
corresponding to this change in @xmath124 , we have @xmath125 \ ; , \label{eq : recoil3}\end{aligned}\ ] ] where , as earlier , @xmath126 , and the positron velocity becomes @xmath127 \ ! ,
\label{eq : beta}\end{aligned}\ ] ] where @xmath128 .
note that the positron energy , @xmath66 and the velocity , @xmath7 , are equal to the recoil - corrected @xmath129 and @xmath130 in ref .
@xcite , respectively . reflecting these changes , the phase space integral in eq.([eq : nlo ] )
needs to be corrected as follows : @xmath131 \ , , \label{eq : recoil2}\end{aligned}\ ] ] where @xmath132 .
the factor @xmath133 in eq.([eq : recoil2 ] ) has corrections of order @xmath134 , and the jacobian factor produces the following nlo phase space factor in eq.([eq : nlo ] ) : @xmath135 \ ; , \label{eq : ps - recoil}\end{aligned}\ ] ] where the expressions for @xmath66 and @xmath7 are given in eqs .
( [ eq : recoil3 ] ) and ( [ eq : beta ] ) .
* corrections from the next - to - leading - order lagrangian * + the @xmath9 corrections to the lagrangian are explicitly written in eq.([eq : lnnee ] ) . as noted before , the radiative corrections to these additional terms in the lagrangian are of higher order than nlo in our counting and hence need not be considered here . evaluating the nlo lagrangian recoil correction contributions , illustrated in diagram ( a ) in fig .
[ fig : feynmanqed ] , we obtain the recoil terms in eq.([eq : gi ] ) @xmath136 comparing these results with those obtained for neutron @xmath7-decay @xcite , we note that there are several relative sign differences .
rcs which are universal at nlo in effective field theory . ]
apart from the @xmath137 phase - space corrections in neutron @xmath7-decay , the @xmath137 corrections ( arising from the lagrangian @xmath137 interaction terms ) relevant to the neutron life - time are contained in the @xmath138 factor in eq.(14 ) of ref .
@xcite . noting that in neutron beta - decay @xmath139 , we may rewrite the @xmath138 factor as @xmath140 comparison of @xmath138 with @xmath141 in eq.([eq : l - recoil ] ) clearly indicates that the @xmath9 recoil corrections are not identical for the neutron @xmath7-decay and the inverse @xmath7-decay .
moreover , since the weak - magnetism term is dominant , the difference between @xmath138 and @xmath141 are of the same magnitude as the corrections themselves . combining the @xmath9 jacobian factor in the square brackets in eq .
( [ eq : ps - recoil ] ) , and the recoil correction arising from the lagrangian , eq .
( [ eq : l - recoil ] ) , we confirm the recoil corrections given in eqs .
( 12 ) and ( 13 ) in ref
. @xcite .
we prefer to keep these two @xmath9 corrections separate since one is of a kinematical origin ( phase space correction ) , whereas the other is of a dynamical origin arising from the transition matrix element .
in this paper we have derived a model - independent expression for the radiative and @xmath9 corrections for the low - energy anti - neutrino proton reaction to next - to - leading - order in an effective field theory approach .
we have shown that short - distance physics not probed in this low - energy reaction can be subsumed into a single low - energy constant @xmath99 . in the @xmath11-expansion scheme adopted here ,
the @xmath74 and @xmath142 corrections are considered to be of the same order for the reactor anti - neutrino energy range .
we have found that the @xmath9 corrections appearing in eq .
( [ eq : l - recoil ] ) , which originate from the lagrangian eq .
( [ eq : lnnee ] ) , are different from the @xmath9 corrections found in neutron @xmath7-decay , see e.g. , ref . @xcite . therefore , to the order under consideration , it is not advisable to write the inverse beta - decay cross section ( or the positron yield ) in terms of the neutron mean life @xmath10 , as advocated in ref .
@xcite .
the short - distance hadronic physics associated with the lec , @xmath99 , was extensively discussed in refs .
the processes involved in @xmath99 were studied from an effective field theoretical perspective in ref .
@xcite . in principle
, we should be able to determine the lec , @xmath99 , from high - precision experimental data . relegating this determination to future study
, we choose here to estimate @xmath99 at the scale @xmath97 = @xmath143 by comparing the short - distance radiative corrections calculated in the s - m approach @xcite and the expressions obtained in eft in ref .
the result is @xmath144 where , for the sake of definiteness , the value of the axial matching mass @xmath145=@xmath146 gev has been used although its value involves uncertainty @xcite . with this value of @xmath147 , the correction term involving lec in eqs .
( [ eq : g1 ] ) and ( [ eq : g2 ] ) is estimated to be @xmath148 . the dominant first term in eq .
( [ eq : lec ] ) arises from well - known additional box diagrams with z - exchange , replacing the photon - exchange , in electro - weak theory @xcite .
this electro - weak physics can be naturally included in our approach .
however , for an easy comparison with the neutron beta - decay radiative corrections evaluated in ref .
@xcite , we prefer to keep this contribution in the above lec . as for the last two terms in eq .
( [ eq : lec ] ) , we remark that @xmath149 involves genuine short - distance hadron - structure physics , whereas the constant @xmath150 arises from photon - loop diagrams in which the photon couples to the nucleon magnetic moments and also from the hadronic form factors .
the long - range parts of these corrections are naturally included in eft at higher orders than considered in this paper . as a final comment we note that in our work we have used the value of the fermi constant @xmath151 determined from the muon lifetime measurement .
the theoretical expression for @xmath151 is evaluated in standard electroweak theory , and it naturally includes log - terms involving @xmath152 .
these log - terms appear in our expression for @xmath99 , eq . ( [ eq : lec ] ) , and was also considered in ref .
@xcite , see e.g. , refs .
@xcite for details . in summary
the integrated cross section for reaction ( [ eq : reaction ] ) is @xmath153 \right\ } \label{eq : sigma}\end{aligned}\ ] ] where as before , @xmath154 and @xmath155 , and where all @xmath137 corrections in eq.([eq : sigma ] ) except @xmath156 of eq.([eq : l - recoil ] ) originate from the phase - space factor @xmath68 , eq.([eq : ps - recoil ] ) .
* acknowledgements * we are grateful to v. gudkov and t. kubota for useful discussion .
this work is supported in part by the national science foundation grants , phys-0758114 and phy-1068305 .
we use dimensional regularization to isolate the ir singularity . the 3-dimensional integral over @xmath157 in eq.([eq : sigmabr ] )
is replaced with a @xmath158 dimensional integral where @xmath159 for the purpose of handling the ir singularity , i.e. eq.([eq : sigmabr ] ) is rewritten as @xmath160 where @xmath161 .
we note that in dimensional regularization , the angular integration @xmath162 yields ( @xmath163 ) @xmath164 } { [ 1-\tilde{\beta } { \rm cos}(\theta_k ) ] ^2 } } \nonumber \\ & = & \frac{\mu^{4-d}}{8\pi^3 } \left\ { \big [ 1\!+\ ! |\epsilon | \left(\gamma_e \!-\ ! { \rm ln}(4\pi ) \right ) \big ] \left [ -\frac{4}{\tilde{\beta}^2 } + \frac{2}{\tilde{\beta}^3}{\rm ln}\left(\frac{1\!+\!\tilde{\beta}}{1\!-\!\tilde{\beta}}\right ) \right ] + \frac{4 |\epsilon |}{\tilde{\beta}^2 } \ { \cal c}(\tilde{\beta } ) + { \cal o}(\epsilon^2)\right\ } \ ; , \nonumber\end{aligned}\ ] ] where @xmath128 and the function @xmath165 is given by ( see e.g. refs .
@xcite ) @xmath166 \nonumber\\ & & \,\,+\,\,\,2\,{\rm ln}2 \left [ \frac{1}{2\tilde{\beta}}{\rm ln}\!\left(\frac{1+\tilde{\beta}}{1-\tilde{\beta}}\right ) -1\right ] + \frac{1}{\tilde{\beta } } l\!\left(\!\frac{2\tilde{\beta}}{1+\tilde{\beta}}\!\right)\ , , \end{aligned}\ ] ] and @xmath167 is the spence function @xmath168 the integral over the photon momentum @xmath169 exhibit the ir - singularity . when we combine our expression for the integrated bremsstrahlung cross section with the contributions from the virtual photon loops , we find that the ir singularity is removed as it should . the chooz collaboration , m. apollonio _
j. * c 27 * , 331 ( 2003 ) .
the first results from this collaboration can be found at http://doublechooz.in2p3.fr/public/public.php and y. abe _ et al .
_ ( double chooz collaboration ) , arxiv:1112.6353 .
daya bay collaboration , x. guo _
et al . _ , arxiv : hep - ex/0701029 doe proposal ( 2007 ) ; f.p . an _
( daya bay collaboration ) , arxiv:1203.1669 ; see also the home page of this experimental collaboration http://dayabay.bnl.gov reno collaboration , j.k .
_ , arxiv:1003.1391 [ hep - ex ] technical design report ( 2010 ) .
a web page of this experimental collaboration can be found at http://hcpl.knu.ac.kr/neutrino/neutrino.html | for the low - energy anti - neutrino reaction , @xmath0 , which is of great current interest in connection with on - going high - precision neutrino - oscillation experiments , we calculate the differential cross section in a model - independent effective field theory ( eft ) , taking into account radiative corrections of order @xmath1 . in eft , the short - distance radiative corrections are subsumed into well - defined low - energy constants the values of which can in principle be determined from the available neutron beta - decay data . in our low - energy eft ,
the order-@xmath1 radiative corrections are considered to be of the same order as the nucleon recoil corrections , which include the weak magnetism " contribution .
these recoil corrections have been evaluated as well .
we emphasize that eft allows for a systematic evaluation of higher order corrections , providing estimates of theoretical uncertainties in our results .
+ 0.5 cm u. raha , f. myhrer and k.kubodera 0.5 cm department of physics and astronomy , university of south carolina , columbia , sc 29208 |
in the last few years , the bcs - bec crossover in two - component fermi superfluids has become a central topic in ultracold atom physics@xcite .
this crossover is of especial interest since the superfluidity continuously changes from the weak - coupling bcs - type to the bose - einstein condensation ( bec ) of tightly bound cooper pairs , as one increases the strength of a pairing interaction@xcite .
thus the bcs - bec crossover enables us to study fermion superfluidity and boson superfluidity in a unified manner .
the superfluid density @xmath0 is a fundamental quantity which describes the response of a superfluid which arises from a bec@xcite .
the superfluid density was first introduced by landau as part of the two - fluid theory of superfluid @xmath4he@xcite . at @xmath5
, the value of @xmath0 always equals the total carrier density @xmath6 .
( in the bcs - bec crossover , @xmath6 is the total number density of fermions . )
this property is satisfied in both the fermi superfluids and bose superfluids , irrespective of the strength of the interaction between particles .
this is quite different from what is called the condensate fraction @xmath3 , which describes the number of the bose - condensed particles@xcite .
for example , in superfluid @xmath4he , only about 10% of atoms are bose - condensed even at @xmath5 , due to the strong repulsion between the @xmath4he atoms [ for a review , see ref . @xcite . ] .
in contrast , all the atoms contribute to the superfluid density at @xmath5 , namely @xmath7 . in a companion paper , we have discussed some analytical results for the superfluid density @xmath0 in the bcs - bec crossover regime of a uniform superfluid fermi gas@xcite .
going past the weak - coupling bcs theory , we derived an expression for @xmath0 in the gaussian fluctuation level in terms of the fluctuations in the cooper channel .
the resulting expression for the normal fluid density @xmath8 is given by the sum of the usual bcs normal fluid density @xmath9 and a bosonic fluctuation contribution @xmath10 .
while the superfluid density from fermions dominates in the weak - coupling bcs regime , the bosonic fluctuation contribution @xmath10 becomes dominant in the strong - coupling bec regime .
since @xmath10 is absent in the mean - field bcs theory , inclusion of fluctuations in the cooper channel is clearly essential in considering the superfluid density in the bcs - bec crossover . in ref .
@xcite , our expression for @xmath0 was obtained using the thermodynamic potential in the presence of a superfluid flow . in the present paper
, we derive a second expression for @xmath0 by calculating the effect of pairing fluctuations on the single - particle green s function with a supercurrent . in this paper
, we use this expression to numerically calculate @xmath0 in the entire bcs - bec crossover regime at finite temperatures .
however , it can be shown that this expression for @xmath0 ( given in sec .
iii ) is equivalent to the result derived in ref .
@xcite . in calculating the superfluid order parameter @xmath1 as well as fermi chemical potential @xmath2
, we include the effect of the pairing fluctuations following the approach given in ref .
this self - consistent treatment of @xmath1 and @xmath2 is crucial in calculating @xmath0 as a function of the temperature in the bcs - bec crossover .
besides the superfluid density , we also calculate the condensate fraction @xmath3 describing the number of bose - condensed particles@xcite .
@xmath3 is of special interest in superfluid fermi gases , since it can be observed experimentally .
indeed , a finite value of @xmath3 is the signature of the bcs - bec superfluid phase@xcite . in this paper
, we show that strong - coupling pair fluctuations have little effect on @xmath3 in the bcs - bec crossover .
we note that @xmath3 has been recently calculated at @xmath5 within a simple mean - field bcs approach@xcite , and using monte carlo techniques@xcite . in this paper , we present detailed results for @xmath3 at finite temperatures in the bcs - bec crossover , based on the nsr theory of fluctuations@xcite .
the present paper is organized as follows . in sec .
ii , the bcs green s functions are solved numerically for the superfluid order parameter @xmath1 and chemical potential @xmath2 self - consistently .
this is done for the entire bcs - bec crossover region and at finite temperatures . in sec .
iii , we calculate the superfluid density @xmath0 treating the strong - coupling pair fluctuation effects within a gaussian approximation@xcite .
numerical results for @xmath0 are presented in sec .
iv . in sec .
v , we define and calculate the condensate fraction @xmath3 . throughout this paper , we take @xmath11 .
we also set the volume @xmath12 , so that the number of atoms @xmath13 and the number density @xmath6 are the same .
in superfluid fermi gases , current experiments make use of a broad feshbach resonance to tune the magnitude and sign of the pairing interaction@xcite . in this case
, the superfluid properties can be studied by using the single - channel bcs model , described by the hamiltonian , @xmath14 here , @xmath15 is a creation operator of a fermi atom with pseudo - spin @xmath16 ( which describe the two atomic hyperfine states . ) .
the fermi atoms have kinetic energy @xmath17 , measured from the fermi chemical potential @xmath2 .
@xmath18 describes a pairing interaction between different fermi atoms .
the magnitude and sign of @xmath19 can be tuned using the feshbach resonance using an external magnetic field .
the weak - coupling bcs limit corresponds to @xmath20 .
we only consider a uniform gas in this paper .
nozires and schmitt - rink ( nsr ) first discussed the bcs - bec crossover behavior of the hamiltonian in eq .
( [ eq.1 ] ) to determine @xmath21@xcite , which we briefly summarize ( see also ref .
@xcite ) , based on a gaussian approximation for pair fluctuations@xcite .
the nsr theory gives two coupled equations , one for @xmath21 , and one giving @xmath2 as a function of @xmath13 .
the equation for @xmath21 , given by the thouless criterion , has the same form as the weak - coupling bcs gap equation at @xmath21 , @xmath22 where @xmath23 . in the weak - coupling bcs theory
, one finds that the equation of state gives @xmath24 , where @xmath25 is the fermi energy .
however , this result for @xmath2 is not valid in the bcs - bec crossover regime where we must include strong - coupling effects due to pairing fluctuations . in the nsr theory , this deviation
is determined by solving the equation for the number of fermions@xcite , within a @xmath26-matrix approximation in the particle - particle channel , @xmath27 .
\label{eq.6}\ ] ] here @xmath28 is the fermi distribution function , and @xmath29 describes non - interacting fluctuations in the cooper channel , @xmath30 here , @xmath31 is the boson matsubara frequency describing bosonic fluctuations . the second term in eq . ( [ eq.6 ] ) gives the number of cooper pairs which are formed , and which become stable as we go through the feshbach resonance . in the extreme bec limit , all the fermions form such stable molecules .
the nsr theory to the superfluid phase has been studied below @xmath21@xcite .
we briefly review this approach .
it is convenient to introduce the two - component nambu field operator@xcite , @xmath32 in this standard nambu representation , eq .
( [ eq.1 ] ) can be rewritten as@xcite , @xmath33\psi_{\bf p } -{u \over 4}\sum_{\bf q}[\rho_{1,{\bf q}}\rho_{1,-{\bf q}}+\rho_{2,{\bf q}}\rho_{2,-{\bf q } } ] .
\label{eq.3}\end{aligned}\ ] ] the mean - field bcs approximation is described by the first three terms of eq .
( [ eq.3 ] ) , with the @xmath34-matrix nambu - gorkov propagator@xcite @xmath35 where @xmath36 is the fermion matsubara frequency associated with the fermions and @xmath37 are the pauli operators .
@xmath38 describes bogoliubov single - particle excitations , associated with the breaking of a cooper pair .
the order parameter @xmath39 ( taken to be real ) corresponds to an off - diagonal mean - field and is related to the thermal green s function @xmath40 as @xmath41 substituting eq .
( [ eq.9 ] ) into eq .
( [ eq.8 ] ) , we obtain the well - known gap equation , @xmath42 equation ( [ eq.10 ] ) is an implicit equation determining @xmath1 as a function of @xmath2 and @xmath43 . in the path integral formalism , the result in eq .
( [ eq.10 ] ) emerges as the saddle point solution of eq .
( [ eq.3])@xcite . equation ( [ eq.10 ] ) reduces to eq .
( [ eq.5 ] ) when @xmath44 , which defines @xmath21 .
@xmath45 ( @xmath46 ) represent the generalized density operators , defined by@xcite @xmath47 writing out the operators , one sees that @xmath48 and @xmath49 describe amplitude and phase fluctuations of the cooper - pair order parameter @xmath1 , respectively . as shown in eq .
( [ eq.3 ] ) , the interaction separates into amplitude fluctuations ( @xmath50 ) and phase fluctuations ( @xmath51 ) . .
the solid line describes the single particle matrix green s function @xmath52 , and the dashed line represents the pairing interaction @xmath18 .
@xmath53 ( @xmath54 ) are the generalized correlation functions , given by eqs .
( [ eq.16])-([eq.18 ] ) .
[ fig1 ] , width=566,height=113 ] the chemical potential @xmath2 is determined by the equation for the total number of particles @xmath13 .
this equation is conveniently obtained from the thermodynamic identity @xmath55 .
we calculate the thermodynamic potential @xmath56 perturbatively , taking into account the interaction in eq .
( [ eq.3 ] ) . in the nambu representation ,
the fluctuation contribution to @xmath56 is diagrammatically described by fig .
[ fig1 ] . as discussed in refs .
@xcite , one finds [ see appendix a. ] @xmath57 , \label{eq.23}\end{aligned}\ ] ] where the first term @xmath58 \label{eq.24}\ ] ] comes from the mean - field thermodynamic potential @xmath59 describing bcs fermi quasiparticles . the second term in eq .
( [ eq.23 ] ) describes the thermodynamic potential contribution @xmath60 from bosonic collective pair fluctuations .
@xmath61 is a @xmath34-matrix correlation function defined in eq .
( [ eq.20 ] ) .
we note that the @xmath2-derivative in eq .
( [ eq.23 ] ) only acts on the chemical potential in the kinetic energy @xmath62 . equation ( [ eq.23 ] ) reduces to the nsr result in eq .
( [ eq.6 ] ) at @xmath21 [ where @xmath44 and @xmath63 .
see appendix a for more details .
we solve eq .
( [ eq.10 ] ) together with eq .
( [ eq.23 ] ) to give @xmath1 and @xmath2 self - consistently . as usual , we need to introduce a high energy cutoff @xmath64 in these coupled equations . this cutoff can be formally eliminated by introducing the two - body @xmath65-wave scattering length @xmath66@xcite ,
@xmath67 using @xmath66 in place of @xmath19 , one can rewrite eqs .
( [ eq.10 ] ) and ( [ eq.23 ] ) in the form @xmath68 , \label{eq.26}\end{aligned}\ ] ] @xmath69 \bigr ] , \label{eq.27}\end{aligned}\ ] ] where the momentum sums are now no longer divergent .
the weak - coupling bcs regime and the strong - coupling bec regime are , respectively , given by @xmath70 and @xmath71 ( where @xmath72 is the fermi momentum ) .
the region @xmath73 is referred to as the crossover regime . " in appendix b , we show that eqs .
( [ eq.26 ] ) and ( [ eq.27 ] ) are the limiting results obtained from a two - channel coupled fermion - boson model@xcite when the feshbach resonance is broad . , and ( b ) fermi chemical potential @xmath2 in the bcs - bec crossover .
the pairing interaction is measured in terms of the inverse of the two - body scattering length @xmath66 , normalized by the fermi momentum @xmath72 . in these panels ,
the dotted line shows @xmath21 as a function of @xmath74 . in the strong - coupling regime ,
the apparent first order behavior of the phase transition is an artifact of the nsr gaussian treatment of pairing fluctuations ( see text ) .
[ fig2 ] , width=302,height=377 ] , and ( b ) chemical potential @xmath75 in the bcs - bec crossover . in panel
( b ) , @xmath2 at @xmath5 is also shown .
the curve ` bec ' gives the strong - coupling bec limit , where one finds @xmath76 .
[ fig3 ] , width=302,height=377 ] as a function of temperature . ` bcs '
labels the weak - coupling bcs limit .
the bendover near @xmath21 is an artifact of our nsr gaussian treatment of fluctuations .
[ fig4 ] , width=377,height=226 ] as a function of temperature . for comparison , in the unitarity limit [ @xmath77 , mc results@xcite
give @xmath78 , and an improved version of nsr theory@xcite gives @xmath79 just below @xmath21 .
[ fig5 ] , width=377,height=226 ] figure [ fig2 ] shows our self - consistent solutions of the coupled equations ( [ eq.26 ] ) and ( [ eq.27 ] ) in the bcs - bec crossover at finite temperatures .
these reproduce the nsr results for @xmath21@xcite , as shown explicitly in fig .
[ fig3 ] . when one enters the crossover region , fig .
[ fig4 ] shows that the order parameter @xmath1 deviates from the weak - coupling bcs result . in the strong - coupling bec regime
, although the superfluid phase transition approaches the value @xmath80@xcite , @xmath81 continues to increase . as a result , the ratio @xmath82 in the bec regime is larger than the weak - coupling bcs universal constant @xmath83 .
we recall that on the bec side of the crossover , where @xmath2 is negative , the energy gap is not equal to @xmath1@xcite .
the chemical potential is strongly affected by fluctuations in the cooper channel , and becomes negative in the bec regime , as shown in figs .
[ fig2](b ) and [ fig3](b ) . in the strong - coupling bec regime ,
@xmath2 approaches @xmath76@xcite .
although @xmath2 strongly depends on the magnitude of the interaction , fig .
[ fig5 ] shows that the temperature dependence of @xmath2 is very weak in the entire bcs - bec crossover .
our results in the unitarity limit are in quite good agreement with quantum monte carlo simulation@xcite as well as a more self - consistent version of nsr@xcite . in fig .
[ fig2 ] , the apparent first - order phase transition in the bec regime is an artifact of the approximate nsr theory we are using .
the reason is as follows . in the bec regime ( where @xmath84 ) , the single - particle bcs excitations @xmath38 have a large energy gap given by @xmath85
this energy gap still exists at @xmath21 , where @xmath44 . in this regime
, we can set @xmath86 in eq .
( [ eq.10 ] ) and @xmath87 in eqs .
( [ eq.16])-([eq.18 ] ) . then , eq . ( [ eq.26 ] ) reduces to the expression @xmath76 , and eq . ( [ eq.27 ] ) becomes , by expanding the correlation functions @xmath88 in powers of @xmath89 and @xmath90 [ for the details , see a similar calculation given in ref .
] , @xmath91 \nonumber \\ & \equiv & n_{c0}+n_d .
\label{eq.36}\end{aligned}\ ] ] here , @xmath92 ( @xmath93 ) and @xmath94 is the bogoliubov excitation spectrum in an interacting gas of bose molecules .
@xmath95 in eq .
( [ eq.36 ] ) is the bose green s function , describing bogoliubov excitations @xmath96 where @xmath97 ( where in our theory @xmath98 , and @xmath93 ) is the effective @xmath65-wave repulsive interaction between cooper pairs .
@xmath99 is given by @xmath100 in sec .
iv , we prove that @xmath99 as defined in eq .
( [ eq.38 ] ) corresponds precisely with the formal definition for the condensate fraction in a fermi superfluid in the bec limit . @xmath101 as defined in eq .
( [ eq.36 ] ) is the number of molecules which are not bose - condensed , again in the bec limit .
equations ( [ eq.36 ] ) and ( [ eq.37 ] ) show that the strong - coupling bec limit corresponds to the popov approximation for a weakly - interacting molecular bose gas@xcite . as is well known ( see , for example , ref .
@xcite ) , the popov approximation gives a spurious first - order phase transition at @xmath21 .
this is the origin of the bendover or first - order phase transition evident in fig .
[ fig2]@xcite and other figures in this paper .
it is well known how to overcome this problem , namely one has to include many - body renormalization effects due to the interaction @xmath102@xcite .
including such higher order corrections past the nsr gaussian fluctuations considered in this paper is also crucial for determining the correct value of effective interaction @xmath102 .
the molecular scattering length @xmath98 which is obtained in eq .
( [ eq.36 ] ) is characteristic of the nsr treatment of fluctuations@xcite .
the correct result @xmath103@xcite requires going past the gaussian approximation@xcite . in this paper , in contrast , we only treat pairing fluctuations within nsr .
however , within this approximation , we calculate the superfluid density ( and condensate fraction @xmath3 in sec v ) in a consistent manner . as discussed in ref .
@xcite , the popov approximation becomes invalid in the small region close to @xmath21 given by @xmath104 although we plot numerical results in the present paper in the whole temperature region for completeness , we emphasize that the restriction in eq .
( [ add2 ] ) also holds in the bec regime .
we note that the region defined in eq .
( [ add2 ] ) becomes narrow as one enters deeper into the bec regime , simply because the molecular scattering length @xmath105 becomes small .
thus , one obtains @xmath106 at @xmath107 ( the case shown in fig .
[ fig4 ] , for example ) but @xmath108 at @xmath109 . as fig .
[ fig4 ] shows , the bendover occurs over an increasingly small region as we go deeper in the bec region .
although eq .
( [ eq.36 ] ) was obtained in the strong - coupling bec regime , we note that the condensate fraction @xmath110 in eq .
( [ eq.38 ] ) is the mean - field approximation @xmath110 for a fermi superfluid .
this is consistent with the result found in sec .
v that the mean - field expression for the condensate fraction @xmath3 is a good approximation even in the strong - coupling bec regime ( at least within nsr ) . the number of molecules @xmath101 in the _ non - condensate _ given in eq .
( [ eq.36 ] ) , in contrast , is due to the pairing fluctuations .
this is discussed in more detail in sec .
in ref . @xcite , our discussion of the superfluid density @xmath0 started from the thermodynamic potential @xmath111 in the presence of an imposed superfluid velocity ( or phase twist ) @xmath112 .
this is given by a simple generalization of @xmath56 given in eq .
( [ eq.19 ] ) , which is for the case @xmath113 . here
we give an alternative formulation of @xmath0 in terms of how the single - particle green s function is altered in the presence of a supercurrent .
our numerical calculations in sec .
iv are based on this expression , but it can be proven to be equivalent to the one discussed in ref .
the expression for @xmath0 discussed in ref .
@xcite is convenient for understanding the role of collective pairing fluctuations .
the result we obtain in this section gives further insight and is convenient for numerical calculations .
when a supercurrent flows in the @xmath114-direction with the superfluid velocity @xmath115 , the supercurrent density @xmath116 is given by @xmath117 .
\label{eq.3.1}\end{aligned}\ ] ] here @xmath118 is the @xmath34-matrix single - particle thermal green s function in the presence of @xmath119 . when the second term in eq .
( [ eq.3.1 ] ) is expanded to @xmath120 , it can be written as @xmath121 , where the superfluid density @xmath0 is defined by @xmath122_{q_z\to 0}\equiv n-\rho_n . \label{eq.3.2}\end{aligned}\ ] ] the second line defines the normal fluid density @xmath123 .
we first evaluate the second term in eq .
( [ eq.3.2 ] ) in the lowest order mean - field approximation . in this case , the supercurrent state is described by the bcs - type hamiltonian @xmath124 \psi_\sigma({\bf r } ) -\int d{\bf r } [ \delta({\bf r } ) \psi^\dagger_\uparrow({\bf r})\psi^\dagger_\downarrow ( { \bf r } ) + h.c . ] , \label{eq.3.3}\ ] ] where @xmath125 is a fermion field operator , and @xmath126 is the superfluid order parameter for the current carrying state . in momentum space ,
( [ eq.3.3 ] ) can be written as @xmath127 \nonumber \\ & = & \sum_{\bf p } { \tilde \psi}_{\bf p}^\dagger \big[{\tilde \xi}_{\bf p } \tau_3+\alpha_{\bf p}-\delta\tau_1 \bigr ] { \tilde \psi}_{\bf p } , \label{eq.3.5}\end{aligned}\ ] ] where constant terms have been left out [ see eq . ( [ eq.3 ] ) ] .
the effects of the supercurrent @xmath119 appear in the doppler shift term , @xmath128 , and in @xmath129 , with @xmath130 .
however , we do not have to take this dependence of @xmath131 on @xmath2 into account in calculating @xmath123 in eq .
( [ eq.3.2 ] ) because it is of order of @xmath132 .
@xmath133 is the two - component nambu field in the supercurrent state , @xmath134 the bcs - gorkov hamiltonian in eq .
( [ eq.3.5 ] ) is described by the @xmath34-matrix single - particle thermal green s function with @xmath135 which has the well - known form [ compare with eq .
( [ eq.9 ] ) ] @xmath136 substituting eq .
( [ eq.3.7 ] ) into eq .
( [ eq.3.2 ] ) , one obtains the mean - field result @xmath137 this is the expected landau expression for the normal fluid density due to fermi quasiparticles with the bcs spectrum @xmath138 when @xmath139 .
this expression gives the fermi contribution @xmath9 to the total normal fluid density @xmath123 throughout the bcs - bec crossover , and is not limited to the weak - coupling bcs limit .
see , ref .
@xcite for additional discussion .
in addition to @xmath9 as given by eq .
( [ eq.46 ] ) , the correction to @xmath140 to first order in @xmath119 gives rise to an additional fluctuation contribution @xmath10 to the normal fluid density .
this should be consistent with the number equation in eq .
( [ eq.23 ] ) , for the @xmath139 state , @xmath141 , \label{eq.3.8}\end{aligned}\ ] ] where the renormalized single - particle green s function @xmath142 @xmath143 involves the correction from the lowest order quasiparticle self - energy due to coupling with collective fluctuations @xmath144 .
\label{eq.3.10}\end{aligned}\ ] ] here , @xmath145 $ ] , and @xmath146 . in obtaining this result ,
we have carried out the @xmath2-derivative on @xmath147 in eq .
( [ eq.23 ] ) by using the identity , @xmath148 for example , @xmath149 \nonumber \\ & + & { \rm tr}[\tau_3{\hat g}_0({\bf p}+{\bf q},i\omega_m+i\nu_n)\tau_+{\hat g}({\bf p},i\omega_m)\tau_-{\hat g}_0({\bf p}+{\bf q},i\omega_m+i\nu_n ) ] \bigr ] .
\label{eq.3.8z}\end{aligned}\ ] ] in the presence of a supercurrent , the green s function analogous to eq .
( [ eq.3.9b ] ) is given by @xmath150 where @xmath151 is given by eq .
( [ eq.3.10 ] ) but with @xmath140 in eq .
( [ eq.3.7 ] ) replacing @xmath52 . substituting eq .
( [ eq.3.11 ] ) into eq .
( [ eq.3.2 ] ) , we obtain @xmath152 the fermi contribution is given by eq .
( [ eq.46 ] ) .
the bosonic fluctuation contribution @xmath10 is given by @xmath153_{q_z\to 0 } \nonumber \\ & = & -{2 \over \beta}\sum_{{\bf p},\omega_m}p_z{\partial \over \partial q_z } { \rm tr } \bigl [ { \hat \sigma}({\bf p},i\omega_m ) { \partial { \hat g}_0({\bf p},i\omega_m ) \over \partial i\omega_m } \bigr]_{q_z\to 0 } , \label{eq.3.15}\end{aligned}\ ] ] where we have used the identity @xmath154 substituting eq . ( [ eq.3.10 ] ) [ with @xmath155 into eq . ( [ eq.3.15 ] ) , and using the identity @xmath156 we find @xmath157\tau_+{\hat g}_0({\bf p}_+,i\omega_m+i\nu_n)\tau_-{\partial { \hat g}_0({\bf p}_-,i\omega_m ) \over \partial i\omega_m } \nonumber \\ & + & [ 1+u\xi_{22}({\bf q},i\nu_n)]\tau_-{\hat g}_0({\bf
p}_+,i\omega_m+i\nu_n)\tau_+{\partial { \hat g}_0({\bf p}_-,i\omega_m ) \over \partial i\omega_m } \nonumber \\ & - & 2u \xi_{12}({\bf q},i\nu_n)\tau_+{\hat g}_0({\bf p}_+,i\omega_m+i\nu_n ) \tau_+{\partial { \hat g}_0({\bf
p}_-,i\omega_m ) \over \partial i\omega_m } \bigr]_{q_z\to 0 } \nonumber \\ & = & - { 2 m \over \beta } { \partial \over \partial q_z } \sum_{{\bf q},\nu_n } { u \over \eta({\bf q},i\nu_n ) } \bigl [ [ 1+u\xi_{11}({\bf q},i\nu_n ) ] { \partial \xi_{22}({\bf q},i\nu_n ) \over
\partial q_z } \nonumber \\ & + & [ 1+u\xi_{22}({\bf q},i\nu_n ) ] { \partial \xi_{11}({\bf q},i\nu_n ) \over \partial q_z } -2u\xi_{12}({\bf q},i\nu_n ) { \partial \xi_{12}({\bf q},i\nu_n ) \over
\partial q_z } \bigr]_{q_z \to 0 } , \label{eq.3.17}\end{aligned}\ ] ] where @xmath158 .
we note that the correlation functions @xmath88 appearing in eq .
( [ eq.3.17 ] ) are defined as in eq .
( [ eq.20 ] ) in terms of the single - particle green s functions @xmath140 in the presence of a supercurrent . in eq .
( [ eq.3.17 ] ) , the @xmath159-derivative only acts on @xmath159 in the doppler shift term @xmath160 in @xmath140 in eq .
( [ eq.3.7 ] ) .
it does _ not _ act on the @xmath159 in the shifted chemical potential @xmath161 . to summarize
, we have shown that the total normal fluid density @xmath123 associated with fermionic and bosonic degrees of freedom is given by the sum of their contributions : @xmath162 { \partial \xi_{22}({\bf q},i\nu_n ) \over \partial q_z } \nonumber \\ & + & [ 1+u\xi_{22}({\bf q},i\nu_n ) ] { \partial \xi_{11}({\bf q},i\nu_n ) \over \partial q_z } -2u\xi_{12}({\bf q},i\nu_n ) { \partial \xi_{12}({\bf q},i\nu_n ) \over
\partial q_z } \bigr]_{q_z \to 0}. \label{eq.3.18}\end{aligned}\ ] ] we note that the superfluid density @xmath0 can be also obtained from a current correlation function@xcite .
the present derivation based on calculating the change in the single - particle green s function from eqs .
( [ eq.3.2 ] ) and ( [ eq.3.11 ] ) is equivalent to calculating the current correlation function to first order in @xmath119 , taking into account both self - energy and current vertex corrections . in eq .
( [ eq.3.2 ] ) , the @xmath159-derivative of the first term in eq .
( [ eq.3.11 ] ) gives the bare current response function .
the @xmath159-derivative of @xmath140 in the second term in eq .
( [ eq.3.11 ] ) gives the self - energy corrections , while the @xmath159-derivative of @xmath151 gives the vertex corrections .
see also appendix a of ref .
@xcite for further discussion .
in this section , we present numerical results for the superfluid density @xmath0 , starting from the expression given in eq .
( [ eq.3.18 ] ) in sec .
as we have noted earlier , the expression for @xmath0 derived in ref .
@xcite would give identical results .
we emphasize that these numerical results for @xmath0 use the renormalized values of both @xmath1 and @xmath2@xcite which determine the bcs quasiparticle spectrum over the entire bcs - bec crossover .
figure [ fig6 ] shows the calculated superfluid density @xmath0 in the bcs - bec crossover . the spurious first - order behavior near @xmath21 in the strong - coupling regime is also seen in the self - consistent solutions for @xmath1 and @xmath2 in fig .
[ fig2 ] . as discussed in sec .
ii , this behavior near @xmath21 is removed through a more sophisticated treatment of fluctuations , which could lead to the correct second order phase transition .
we plot our calculated results for @xmath0 close to @xmath21 , in spite of this problem .
in particular , we note that the predicted value of @xmath21 in the nsr theory of the bcs - bec crossover is a good approximation@xcite .
the nsr bendover is much less in evidence in the case of a narrow feshbach resonance , as considered in ref .
@xcite . in appendix c , we prove analytically that our expression in eq .
( [ eq.3.18 ] ) gives @xmath163 at @xmath5 and also that @xmath0 vanishes as @xmath164 ( normal phase ) . getting these two limits correctly ( as shown in fig .
[ fig6 ] ) is very important in any theory of the superfluid density . in the bcs - bec crossover .
the self - consistent solutions for @xmath1 and @xmath2 shown in fig .
[ fig2 ] are used .
the dashed line shows @xmath21 [ see fig .
[ fig3](a ) ] .
[ fig6 ] , width=302,height=377 ] as a function of temperature in the bcs region ( solid circles ) , unitarity limit ( solid triangles ) and bec regime ( open circles ) . ` bcs ' labels the mean - field bcs result , given by @xmath165 with @xmath24 . `
bec ' gives @xmath0 for a dilute bose gas with @xmath166 bosons described by the excitation spectrum in eq .
( [ eq.37 ] ) .
[ fig7 ] , width=302,height=226 ] figure [ fig7 ] shows @xmath0 as a function of temperature in the bcs regime , unitarity limit , and the bec regime .
we note that @xmath0 in the bec regime [ @xmath107 ] is in good agreement with the superfluid density @xmath0 of a weakly interacting gas of @xmath166 bose molecules described by bogoliubov - popov excitations in eq .
( [ eq.37 ] ) , as one expects in the extreme bec limit .
more precisely , in this limit one can show ( see ref .
@xcite for details ) that @xmath168 where @xmath10 is given by the landau formula for the normal fluid of an interacting bose gas @xmath169 here , @xmath93 is the cooper - pair mass and @xmath170 is the bose distribution function . the excitation energy @xmath171 is given by eq .
( [ eq.37 ] ) , calculated with the correct values of @xmath1 and @xmath2 .
the curve labeled by bec in fig . [ fig7 ] corresponds to the result obtained using eqs .
( [ eq.bec1 ] ) and ( [ eq.bec2 ] ) .
this shows the importance of a consistent treatment of fluctuation effects in calculating @xmath0 , @xmath1 , and @xmath2 .
as one approaches the weak - coupling bcs regime , pair fluctuations become weak , so that in this limit the fermionic contribution @xmath9 becomes dominant .
thus one has @xmath172 where @xmath9 is given by the landau formula for the normal fluid in eq .
( [ eq.46 ] ) with the bcs quasiparticle energies @xmath138 .
the curve labeled by bcs in fig .
[ fig7 ] corresponds to the result obtained using eqs .
( [ eq.bec3 ] ) and ( [ eq.46 ] ) .
in this section , we calculate the condensate fraction @xmath3 in the bcs - bec crossover .
the condensate fraction @xmath3 in the superfluid phase is most conveniently defined@xcite as the maximum eigenvalue of the two - particle density matrix , @xmath173 , where @xmath125 is a fermion field operator . the condensate fraction @xmath3 is given as the maximum eigenvalue , of order @xmath13 . when only one eigenvalue is @xmath174
, we find @xmath175 where terms of order @xmath176 have been ignored . here
@xmath177 is the ( normalized ) two - particle eigenfunction of @xmath178 with the eigenvalue @xmath3 .
the off - diagonal long range order of a fermi superfluid@xcite is characterized as , for large separation of ( @xmath179 ) and ( @xmath180 ) , @xmath181 comparing eqs .
( [ eq.40 ] ) and ( [ eq.41 ] ) , the condensate fraction @xmath3 is seen to be the normalization factor of the cooper - pair wavefunction , @xmath182 ( see , for example , ref .
@xcite ) , @xmath183 in physical terms , the maximum eigenvalue @xmath3 describes the occupancy of two - particle states . in a uniform fermi superfluid
, the bcs mean - field approximation gives @xmath184 in the strong - coupling bec regime ( where @xmath84 ) , we can set @xmath185 in eq .
( [ eq.43 ] ) . in this case , substituting eq .
( [ eq.43 ] ) into eq .
( [ eq.42 ] ) , the mean - field expression for the condensate fraction ( @xmath186 ) in the bec regime reduces to @xmath187 in obtaining this expression , we have used the fact that @xmath188 in the bec regime@xcite . more generally , in terms of the single - particle green s functions
, one can write eq .
( [ eq.42 ] ) as @xmath189 to calculate the strong - coupling effects on @xmath3 , we substitute eq .
( [ eq.3.9b ] ) into eq .
( [ eq.4.1 ] ) .
since this green s function @xmath190 only includes first order self - energy corrections , we only retain the correction terms to @xmath3 to @xmath191 , giving @xmath192 here , the mean - field component @xmath99 is the bcs fermi quasiparticle contribution @xmath193 the first order fluctuation contribution @xmath194 is given by @xmath195 \nonumber \\
& + & { \rm tr}[\tau_+{\hat g}_0({\bf p},i\omega_m){\hat \sigma}({\bf p},i\omega_m){\hat g}_0({\bf p},i\omega_m ) ] g_0^{12}({\bf p},i\omega_m ' ) \bigr ] .
\label{eq.4.4}\end{aligned}\ ] ] the correction term @xmath194 in eq .
( [ eq.4.4 ] ) is not important in the weak - coupling bcs regime , where fluctuation effects clearly can be ignored .
figure [ fig8 ] shows that @xmath196 is also negligibly small in the strong - coupling regime .
thus @xmath3 is well approximated by the mean - field expression in eq .
( [ eq.4.3 ] ) over the entire bcs - bec crossover , at least in our nsr - type approximation .
we recall that the same pair fluctuations made a large contribution to @xmath0 as we went from the bcs to bec region .
the difference is that by definition , @xmath196 in eq .
( [ eq.4.4 ] ) arises from self - energy corrections to the single - particle anomalous green s function @xmath197 .
there is no distinct bosonic contribution , such as @xmath10 in the normal fluid density .
thus it is not unexpected that the fluctuations are a small correction to @xmath3 .
we note , however , since fluctuations in the cooper channel are taken into account in the equation of state in eq .
( [ eq.27 ] ) , they modify the condensate fraction @xmath3 given by eq .
( [ eq.4.3 ] ) .
for example , let us consider the bec regime at @xmath5 , where the gap equation gives @xmath76 . in this limit , the number equation reduces to @xmath198 where @xmath99 is given by eq .
( [ eq.38 ] ) and [ taking the @xmath199 limit of eq .
( [ eq.36 ] ) ] @xmath200 \simeq { 8 \over 3\sqrt{\pi}}(n_{c0}a_m)^{3 \over 2 } \label{eq.4.6}\ ] ] gives the quantum depletion from the molecular condensate due to the effective interaction @xmath102 between cooper pairs .
recently , the mean - field result in eq .
( [ eq.4.3 ] ) has been used to study the condensate fraction @xmath3 in a superfluid fermi gas at @xmath5@xcite . in this case , the gap and number equation in the mean - field approximation for the bcs - bec crossover reduce to @xmath201 , \label{eq.4.7}\ ] ] @xmath202 .
\label{eq.4.8}\ ] ] in the bec regime , while eq .
( [ eq.4.7 ] ) again gives @xmath76 , eq .
( [ eq.4.8 ] ) reduces to @xmath203 in eq .
( [ eq.36 ] ) .
as expected , the depletion @xmath101 at @xmath5 from the condensate due to the interaction between cooper pairs is omitted when the bcs - bec crossover is described by mean - field approximation@xcite . in the strong - coupling bec regime .
the solid line shows @xmath99 and the dashed line includes the small correction @xmath196 from self - energies due to pairing fluctuations .
[ fig8 ] , width=377,height=226 ] in the bcs - bec crossover at @xmath5 ( solid line ) . in this figure , as well as in figs .
[ fig10 ] and [ fig11 ] , we only show @xmath99 . `
mf ' shows the condensate fraction in the case when @xmath1 and @xmath2 are determined in the mean - field results in eqs .
( [ eq.4.7 ] ) and ( [ eq.4.8 ] ) .
` bec ' shows the result for a bose gas described by eqs .
( [ eq.4.5 ] ) and ( [ eq.4.6 ] ) .
the solid circles shows recent monte carlo results for @xmath3@xcite for comparison .
[ fig9 ] , width=377,height=226 ] figure [ fig9 ] shows the condensate fraction in the bcs - bec crossover regime at @xmath5 . because of the omission of the quantum depletion effect , the simple mean - field result ( mf ) given in ref .
@xcite is larger than our result ( @xmath99 ) . in the region @xmath204 ,
our results are well described by the condensate fraction for a superfluid molecular bose gas given by eqs .
( [ eq.4.5 ] ) and ( [ eq.4.6 ] ) ( labeled bec in fig .
[ fig9 ] ) .
figure [ fig9 ] also compares our @xmath5 results with those obtained by quantum monte carlo ( mc ) simulations@xcite .
the latter calculation gives results consistent with @xmath103@xcite .
in contrast , our nsr theory gives the larger mean - field molecular scattering length @xmath98 .
as a result , we overestimate the magnitude of the depletion and thus our values for @xmath3 are smaller than the mc calculation in the bcs - bec crossover regime .
the measurement of the depletion deep in the bec regime would be a useful way of determining the magnitude of @xmath205 . as a function of temperature in the bcs - bec crossover .
in this calculation , the self - consistent solutions for @xmath1 and @xmath2 in fig . [ fig2 ] are used .
[ fig10 ] , width=377,height=453 ] as a function of temperature in the bcs regime , unitarity limit , and the bec regime .
the curve ` bcs ' is the weakly - coupling bcs result .
the curve labeled ` bec ' is the condensate fraction for a bose superfluid determined by eq .
( [ eq.36 ] ) .
results are normalized to values at @xmath5 .
[ fig11 ] , width=377,height=226 ] figure [ fig10 ] shows the condensate fraction in the bcs - bec crossover at finite temperatures . in the weak - coupling bcs regime ,
the condensate fraction @xmath3 is very small even far below @xmath21 , because only atoms very close to the fermi surface form cooper pairs which are bose - condensed . in this regime ,
[ fig11 ] shows that the temperature dependence of @xmath3 is very well described by the weak - coupling bcs result . in the crossover region ,
the temperature dependence of @xmath3 deviates from the simple bcs result , as shown by the case @xmath206 in fig .
[ fig11 ] . in the bec limit ,
[ fig10 ] shows that the condensate fraction at @xmath5 approaches @xmath166 , reflecting the fact that all atoms form cooper pairs which are bose - condensed . in this regime ,
[ fig11 ] shows that the temperature dependence of @xmath3 agrees with the condensate fraction for a dilute bose gas in the popov approximation given by eq .
( [ eq.36 ] ) .
figure [ fig9 ] shows that the bec picture is a very good approximation when @xmath204 .
the fact that @xmath3 agrees with the popov theory for a weakly interacting molecular bose gas shows that the superfluid phase transition in this regime is dominated by the thermal depletion of cooper pair condensate , and not by the dissociation of cooper pairs characteristic of the weak - coupling bcs regime .
in this paper , we have calculated the superfluid density @xmath0 and condensate fraction @xmath3 in the bcs - bec crossover regime of a uniform superfluid fermi gas at finite temperatures .
we have included strong - coupling fluctuation effects on both @xmath0 and @xmath3 within a gaussian approximation .
the same fluctuation effects were also taken into account in calculating the superfluid order parameter @xmath1 and fermi chemical potential @xmath2 in the bcs - bec crossover , within the nsr theory@xcite .
the expression we used to calculate @xmath0 was derived from the single - particle green s function in the presence of supercurrent as given by eq .
( [ eq.3.1 ] ) , which brings in self - energy corrections due to dynamic pair fluctuations . in this paper , we have concentrated on the explicit numerical calculation of @xmath0 within a gaussian approximation . in contrast , our companion paper@xcite uses a different ( but equivalent ) formulation which exhibits the structure of @xmath0 in a more direct fashion , in particular , the relation to collective modes . our result for the normal fluid density in eq .
( [ eq.3.18 ] ) naturally separates into a mean - field part associated with fermions @xmath9 and a bosonic pairing fluctuation contribution @xmath10 . as discussed in ref .
@xcite , @xmath9 is given by the landau excitation formula in eq .
( [ eq.46 ] ) in the whole bcs - bec crossover . in the strong - coupling bec regime
, @xmath9 is negligible and @xmath10 reduces to the landau formula for the normal fluid of a bose gas of tightly bound copper pairs@xcite .
however , in the region near unitarity , @xmath10 is not expected to be given by a landau - type formula because the bosonic pairing fluctuations are strongly damped .
it is in this region that the numerical calculations for @xmath0 reported in this paper are especially useful .
the superfluid density is a fundamental quantity in two - fluid hydrodynamics@xcite and we will use our results in a future study of hydrodynamic modes in the bcs - bec crossover regime of a fermi superfluid at finite temperatures .
in contrast to the superfluid density , the mean - field expression for the condensate fraction @xmath3 is a good approximation even in the strong - coupling bec regime .
the fluctuation contribution to @xmath3 gives rise to the non - condensate component . in the bec regime
, we showed that the fluctuation contribution gives the condensate depletion @xmath101 due to the effective interaction @xmath102 between cooper pairs , which is finite even at @xmath5 . in the bec regime ,
the strong coupling theory presented in this paper reduces to that of a weakly interacting bose gas of molecules , with an excitation spectrum given by the bogoliubov - popov approximation .
this is also the origin of the spurious first - order phase transition our theory exhibits ( see figs .
[ fig6 ] and [ fig7 ] ) .
this is a well - known problem in dealing with dilute bose gases@xcite .
the recovery of the second order phase transition in the entire bcs - bec crossover , as well as the normalized magnitude of the effective interaction between cooper pairs@xcite , both require the inclusion of higher order fluctuations past the nsr gaussian approximation which we have used . in this regard
, we have emphasized that calculating the value of @xmath0 is very dependent on using the strong - coupling approximation for @xmath1 and @xmath2 as well , quantities which determine the single - particle excitation spectrum .
thus , when we calculate @xmath1 and @xmath2 beyond the gaussian fluctuation level , we also need to improve the microscopic model used to calculate @xmath0 .
the approach presented in this paper , as well as in ref .
@xcite , can be the starting point for such improved calculations
. n. f. and y. o. would like to thank h. matsumoto for useful discussions .
y. o. was financially supported by a grant - in - aid for scientific research from the ministry of education , culture , sports , science and technology of japan ( 16740187 , 17540368 , and 18043005 ) .
the work of e. t. and a. g. were supported by nserc of canada .
we briefly review some of the formalism developed in ref . @xcite .
the thermodynamic potential @xmath207 consists of the mean - field part @xmath59 due to fermi quasiparticles and the correction term @xmath60 due to boson fluctuations .
the mean - field contribution @xmath59 is given by@xcite @xmath208 -{2 \over \beta}\sum_{\bf p}\ln [ 1+e^{-\beta e_{\bf p } } ] .
\label{eq.11}\end{aligned}\ ] ] in the nsr gaussian approximation , the fluctuation contribution is the sum of diagrams @xmath60 shown in fig .
[ fig1 ] , namely @xmath209 , \label{eq.19}\end{aligned}\ ] ] where @xmath210 the correlation functions @xmath53 are given within the mean - field approximation by @xmath211 .
\label{eq.14}\end{aligned}\ ] ] @xmath212 and @xmath213 describe the amplitude fluctuations and phase fluctuations of the order parameter , respectively , while @xmath214 and @xmath215 are the coupling of these fluctuations . doing the fermi matsubara frequency sum over @xmath36 , one finds@xcite @xmath216 \nonumber \\ & - & \sum_{\bf p } \left ( 1+{\xi_{{\bf p+q}/2}\xi_{{\bf p - q}/2}-\delta^2 \over e_{{\bf p+q}/2}e_{{\bf p - q}/2 } } \right ) { e_{{\bf p+q}/2}+e_{{\bf p - q}/2 } \over ( e_{{\bf p+q}/2}+e_{{\bf p - q}/2})^2+\nu_n^2 } \nonumber \\ & \times & [ 1-f(e_{{\bf p+q}/2})-f(e_{{\bf p - q}/2 } ) ] , \label{eq.16}\end{aligned}\ ] ] @xmath217 \nonumber \\ & - & \sum_{\bf p } \left ( 1+{\xi_{{\bf p+q}/2}\xi_{{\bf p - q}/2}+\delta^2 \over e_{{\bf p+q}/2}e_{{\bf p - q}/2 } } \right ) { e_{{\bf p+q}/2}+e_{{\bf p - q}/2 } \over ( e_{{\bf p+q}/2}+e_{{\bf p - q}/2})^2+\nu_n^2 } \nonumber \\ & \times & [ 1-f(e_{{\bf p+q}/2})-f(e_{{\bf p - q}/2 } ) ] , \label{eq.17}\end{aligned}\ ] ] @xmath218 \nonumber \\ & - & \sum_{{\bf p } } \left ( { \xi_{{\bf p+q}/2 } \over e_{{\bf p+q}/2}}+{\xi_{{\bf p - q}/2 } \over e_{{\bf p - q}/2 } } \right ) { \nu_{n } \over ( e_{{\bf p+q}/2}+e_{{\bf p - q}/2})^2+\nu_{n}^2 } \nonumber \\ & \times & [ 1-f(e_{{\bf p+q}/2})-f(e_{{\bf p - q}/2 } ) ] \nonumber \\ & = & -\pi_{21}^0({\bf q},i\nu_n ) .
\label{eq.18}\end{aligned}\ ] ] we note that for zero superfluid flow ( @xmath139 ) , the matrix elements @xmath219 of the inverse fluctuation propagator discussed in ref .
@xcite are related to @xmath147 by @xmath220 in contrast to ref .
@xcite , eq .
( [ eq.20 ] ) splits the fluctuations explicitly into phase and amplitude components . in calculating @xmath221
, we note that , in principle , both @xmath222 and @xmath1 depend on @xmath2 .
thus , one needs to calculate @xmath223 .
the first term @xmath224 vanishes by definition , since the gap equation ( [ eq.10 ] ) is the saddle point solution@xcite . as noted in our previous paper@xcite , strictly speaking , the second term @xmath225 is a higher order correction within the nsr gaussian treatment of fluctuations .
thus in a consistent theory built on treating fluctuations to quadratic order , we should omit the term @xmath226 . within this approximation
, we obtain the expression for @xmath13 given in eq .
( [ eq.23 ] ) . if we retained terms involving @xmath225 in eq .
( [ eq.6 ] ) , the resulting equation for @xmath13 would not reduce to the correct result at @xmath21 in the limit @xmath164 . on the other hand ,
the importance of this class of higher order correction terms has been pointed out in refs .
in particular , ref .
@xcite has shown that the correct molecular scattering length in the bec regime ( @xmath103 ) is obtained when this term is included .
@xcite for additional discussion .
in this appendix , we compare the results for a single - channel model given by eqs .
( [ eq.26 ] ) and ( [ eq.27 ] ) with our previous results for the coupled fermion - boson ( cfb ) model@xcite .
the cfb two - channel model involves treating the feshbach resonance explicitly by the hamiltonian , @xmath227 c_{{\bf p}\sigma}^\dagger c_{{\bf p}\sigma } -u_{\rm bg}\sum_{{\bf p},{\bf p}',{\bf q } } c_{{\bf p}+{\bf q}\uparrow}^\dagger c_{{\bf p}'-{\bf q}\downarrow}^\dagger c_{{\bf p}'\downarrow } c_{{\bf p}\uparrow } \nonumber \\ & + & \sum_{\bf q}[\varepsilon_{\bf q}^b+2\nu-2\mu]b_{\bf q}^\dagger b_{\bf q } + g_{\rm r } \sum_{{\bf p},{\bf q } } [ b_{\bf q}^\dagger c_{{\bf p}+{\bf q}/2,\downarrow } c_{-{\bf p}+{\bf q}/2,\uparrow } + h.c . ] .
\label{eq.28}\end{aligned}\ ] ] here , @xmath228 is the creation operator of a molecular boson associated with the feshbach resonance , with the kinetic energy @xmath229 .
the threshold energy @xmath230 of the feshbach resonance can be tuned by adjusting a small external magnetic field .
@xmath231 describes the feshbach coupling between atoms and a molecule .
@xmath232 is a nonresonant weak - interaction , which is taken to be attractive in eq .
( [ eq.28 ] ) .
since one molecule is a bound state of two fermi atoms ( in different hyperfine states ) , we take @xmath93 and must impose a conservation law for the total number of fermi atoms .
the latter constraint has been taken into account in eq .
( [ eq.28 ] ) by taking the chemical potential for the molecules to be @xmath233 . in a previous paper@xcite
, we used this two - channel cfb model to treat the superfluid properties in the bcs - bec crossover region within the nsr theory .
the resulting coupled equations corresponding to eqs .
( [ eq.10 ] ) and ( [ eq.23 ] ) are given by @xmath234 @xmath235{\hat \xi}({\bf q},i\nu_n ) \bigr ] .
\label{eq.30}\end{aligned}\ ] ] here , @xmath236 is an effective pairing interaction associated with the feshbach resonance .
@xmath237 is the bec order parameter and @xmath238 is the number of bose - condensed feshbach molecules . because of the resonance coupling between atoms and molecules , @xmath239 is related to the bcs order parameter @xmath1 by the relation @xmath240@xcite .
@xmath241 gives the thermal occupation of the non - condensed molecules , where @xmath242 is the bose distribution function .
@xmath243 is the @xmath34-matrix single - particle green s function for a free bose gas of molecules . in eq .
( [ eq.29 ] ) , the bcs single - particle excitation spectrum @xmath244 is given by @xmath245 , where @xmath246 and @xmath247 is the composite order parameter involving the @xmath1 and @xmath239 coupled order parameters .
for more details , see ref .
@xcite .
let use now consider the case of a broad feshbach resonance , where the coupling @xmath231 is very large ( @xmath248 ) . in this limit , the effective pairing interaction @xmath249 defined below eq .
( [ eq.30 ] ) can be strong even when the threshold energy @xmath230 is still much larger than the chemical potential @xmath2 ( @xmath250 ) . in this case , one can neglect @xmath2 in @xmath249 in the interesting bcs - bec crossover regime . since @xmath230 is the lowest excitation energy of feshbach molecules , we can also neglect @xmath251 and @xmath252 in eq .
( [ eq.30 ] ) when @xmath253 .
similarly , for @xmath253 , dynamical effects of feshbach molecules are not important , and we can use the approximation @xmath254 in eq .
( [ eq.30 ] ) .
the end result is that the coupled equations ( [ eq.29 ] ) and ( [ eq.30 ] ) for the two - channel model reduce to a form analogous to eqs . ( [ eq.10 ] ) and ( [ eq.23 ] ) in the single - channel model .
we need only replace @xmath1 by @xmath255@xcite and @xmath19 by the two - particle effective interaction associated with the feshbach resonance , @xmath256 thus we see how the single - channel description of the bcs - bec crossover described by eqs .
( [ eq.10 ] ) and ( [ eq.23 ] ) is a special case of the two - channel model in the case of a broad feshbach resonance @xmath253 .
the single - channel scattering length @xmath66 is related to the two - channel cfb model parameters by @xmath257 here , @xmath258 ,
@xmath259 , and @xmath260 are all renormalized quantities , given by@xcite @xmath261 the renormalized unitarity limit ( @xmath262 ) corresponds to @xmath263 .
we first prove that the normal fluid density as given by the expression in eq .
( [ eq.3.18 ] ) vanishes at @xmath5 . at @xmath5 ,
clearly the first term @xmath264 vanishes . for the fluctuation part @xmath10
, we find from explicit calculations that one can change the @xmath159-derivative into the @xmath31-derivative .
then , we find @xmath265 \nonumber \\ & = & -{2 \over \pi}\sum_{\bf q}{q_z^2 \over 2 m } \int_{-\infty}^\infty dz
n_b(z ) { \rm i m } \bigl [ { \partial^2 \over \partial z^2 } \ln { \rm det } [ 1+u{\hat \xi}({\bf q},i\nu_n\to z+i\delta ) ] \bigr ] \nonumber \\ & = & { 2 \over \pi } \sum_{\bf q } { q_z^2 \over 2 m } { \rm i m } \bigl [ { \partial \over \partial z } \ln { \rm det } [ 1+u{\hat \xi}({\bf q},z+i\delta ) \bigr]_{z=-\infty}^{z=0}. \label{eq.3.19}\end{aligned}\ ] ] in obtaining the last line , we have used that @xmath266 at @xmath5 .
one can easily show that eq .
( [ eq.3.19 ] ) vanishes , by noting that @xmath267 and @xmath268 , @xmath269 , and @xmath270 are all real quantities .
thus , we have shown that in our microscopic model , @xmath271 at @xmath5 , or @xmath163 .
we next show that @xmath0 vanishes in the limit @xmath164 , i.e. , in the normal phase above @xmath21 . to see this
, we note that the mean - field part @xmath264 reduces to the number of fermi atoms when @xmath164 , namely , @xmath272 for the fluctuation contribution @xmath10 , since @xmath273 in the limit @xmath164 , eq .
( [ eq.3.17 ] ) reduces to @xmath274 g_0^{22}({\bf p}_+,i\omega_m+i\nu_n ) { \partial g_0^{11}({\bf p}_-,i\omega_m )
\over \partial i\omega_m } \nonumber \\ & + & [ 1+u\xi_{22}({\bf q},i\nu_n ) ] g_0^{11}({\bf p}_+,i\omega_m+i\nu_n ) { \partial g_0^{22}({\bf p}_-,i\omega_m ) \over \partial i\omega_m } \bigr]_{q_z\to 0}. \label{eq.3.21}\end{aligned}\ ] ] carrying out the @xmath159-derivative , we obtain @xmath275 g_0^{22}({\bf p}_+,i\omega_m+i\nu_n ) { \partial g_0^{11}({\bf p}_-,i\omega_m ) \over
\partial i\omega_m } \nonumber \\ & - & [ 1+u\xi_{22}({\bf q},i\nu_n ) ] g_0^{11}({\bf p}_+,i\omega_m+i\nu_n ) { \partial g_0^{22}({\bf p}_-,i\omega_m ) \over \partial i\omega_m } \bigr ] \nonumber \\ & = & { 1 \over \beta^2 } \sum_{{\bf p},\omega_m } \sum_{{\bf q},\nu_n } { u \over \eta({\bf q},i\nu_n ) } [ 1+u\xi_{22}({\bf q},i\nu_n ) ] \nonumber \\ & \times & \bigl [ g_0^{11}({\bf p}_+,i\omega_m+i\nu_n ) { \partial g_0^{22}({\bf p}_-,i\omega_m ) \over \partial i\omega_m } - g_0^{22}({\bf p}_-,i\omega_m ) { \partial g_0^{11}({\bf p}_+,i\omega_m+i\nu_n )
\over \partial i\omega_m } \bigr ] .
\nonumber \\ \label{eq.3.22}\end{aligned}\ ] ] noting that @xmath276[1-u\xi_{22}({\bf q},i\nu_n)]$ ] and @xmath277 when @xmath44 , and using the identity , @xmath278 \nonumber \\ & = & -{\partial \over \partial\mu } { 1 \over \beta}\sum_{{\bf p},\omega_m}[g_0^{11}({\bf p}_+,i\omega_m+i\nu_n ) g_0^{22}({\bf p}_-,i\omega_m ) ] \nonumber \\ & = & { \partial \over \partial \mu}\pi({\bf q},i\nu_n ) , \label{eq.3.23}\end{aligned}\ ] ] one may reduce eq .
( [ eq.3.22 ] ) to @xmath279 .
\label{eq.3.24}\end{aligned}\ ] ] the sum of eqs .
( [ eq.3.20 ] ) and ( [ eq.3.24 ] ) equals the total number of fermi atoms @xmath6 as given in eq .
( [ eq.6 ] ) .
therefore , we have proven that @xmath280 , or @xmath281 in the limit @xmath164 ( normal phase )
. this is an important requirement of any theory of @xmath0 .
99 for a review , see q. chen , j. stajic , s. tan , and k. levin , phys .
rep . * 412 * , 1 ( 2005 ) .
p. nozires and s. schmitt - rink , j. low .
* 59 * , 195 ( 1985 ) .
m. randeria , in _ bose - einstein condensation _ , edited by a. griffin , d. w. snoke and s. stringari ( cambridge university press , n.y . , 1995 ) , p.355 .
y. ohashi and a. griffin , phys .
a * 67 * , 063612 ( 2003 ) . c. regal , m. greiner , and d. jin , phys .
92 * , 040403 ( 2004 ) .
p. nozires and d. pines , _ the theory of quantum liquids _ , vol .
ii ( addison - wesley , ny , 1990 ) , chap . 6 .
l. landau , j. phys .
5 * , 71 ( 1941 ) . c. n. yang , rev .
phys . * 34 * , 694 ( 1962 ) .
a. griffin , _ excitations in a bose - condensed liquid _
( cambridge university press , ny , 1993 ) .
e. taylor , a. griffin , n. fukushima , and y. ohashi , cond - mat/0609187 .
l. salasnich , n. manini , and a. parola , phys .
a * 72 * , 023621 ( 2005 ) .
g. astrakharchik , j. boronat , j. casulleras , and s. giorgini , phys .
95 * , 230405 ( 2005 ) .
j. engelbrecht , m. randeria , and c. s de melo , phys .
b * 55 * , 15153 ( 1997 ) .
y. ohashi and a. griffin , phys .
rev . lett . * 89 * , 130402 ( 2002 ) . in all bose frequency sums in this paper
, one needs the convergence factor @xmath282 .
this is left implicit .
j. schrieffer , _ theory of superconductivity _
( addison - wesley , ny , 1964 ) , chap
k. maki , in _ superconductivity _ , edited by r. d. parks ( marcel dekker , ny , 1969 ) , vol .
2 , p. 1035 . y. ohashi and s. takada , j. phys .
jpn . * 66 * , 2437 ( 1997 ) .
since we take @xmath1 to be real , we need to eliminate the expectation value from @xmath48 as @xmath283 , although this is not explicitly written in eq .
( [ eq.4 ] ) .
a. leggett , in _ modern trends in the theory of condensed matter _ , edited by a. pekalski and j. przystawa ( springer verlag , berlin , 1980 ) , p. 14 .
h. hu , x .- j .
liu , p. drummond , europhys . lett . * 74 * , 574 ( 2006 ) .
e. burovski , n. prokofev , b. svistunov , and m. troyer , phys .
96 * , 160402 ( 2006 ) .
see , for example , c. pethick and h. smith , _ bose - einstein condensation in dilute gases _
( cambridge university press , ny , 2002 ) , chap.8 .
p. pieri and g. strinati , phys .
b * 71 * , 094520 ( 2005 ) . h. shi and a. griffin , phys . rep .
* 304 * , 1 ( 1998 ) .
p. pieri , l. pisani , and g. strinati , phys .
b * 70 * , 094508 ( 2004 ) . for many - body corrections to the interaction between bosons , see , y. ohashi , j. phys .
. jpn . * 74 * , 2659 ( 2005 ) .
see also m. bijlsma , and h. stoof , phys .
a * 54 * , 5085 ( 1996 ) .
d. petrov , c. salomon , and g. shlyapnikov , phys .
lett . * 93 * , 090404 ( 2004 ) .
p. pieri , and g. strinati , phys . rev .
b * 61 * , 15370 ( 2000 ) .
see , for example , chap .
g. baym , in _ mathematical methods in solid state and superfluid theory _ , edited by r. c. clark , and g. h. derrick ( oliver and boyd , edinburgh , 1967 ) p121 .
e. taylor , and a. griffin , phys .
a * 72 * , 053630 ( 2005 ) .
j. keeling , p. eastham , m. szymanska , and p. littlewood , phys .
b * 72 * , 115320 ( 2005 ) . in a broad feshbach resonance , although @xmath239 is small , the feshbach coupling @xmath231 is large . as a result , the molecular component @xmath284 in the composite order parameter @xmath285 still has a relatively large contribution even in the crossover regime . for further discussion , see ref .
@xcite . see , for example , y. ohashi and a. griffin , phys .
a * 67 * , 033603 ( 2003 ) . | the superfluid density is a fundamental quantity describing the response to a rotation as well as in two - fluid collisional hydrodynamics .
we present extensive calculations of the superfluid density @xmath0 in the bcs - bec crossover regime of a uniform superfluid fermi gas at finite temperatures.we include strong - coupling or fluctuation effects on these quantities within a gaussian approximation .
we also incorporate the same fluctuation effects into the bcs single - particle excitations described by the superfluid order parameter @xmath1 and fermi chemical potential @xmath2 , using the nozires and schmitt - rink ( nsr ) approximation .
this treatment is shown to be necessary for consistent treatment of @xmath0 over the entire bcs - bec crossover .
we also calculate the condensate fraction @xmath3 as a function of the temperature , a quantity which is quite different from the superfluid density @xmath0 .
we show that the mean - field expression for the condensate fraction @xmath3 is a good approximation even in the strong - coupling bec regime .
our numerical results show how @xmath0 and @xmath3 depend on temperature , from the weak - coupling bcs region to the bec region of tightly - bound cooper pair molecules . in a companion paper by the authors ( cond - mat/0609187 ) , we derive an equivalent expression for @xmath0 from the thermodynamic potential , which exhibits the role of the pairing fluctuations in a more explicit manner . |
quantum systems are localized in a finite well - defined space area .
this area may be determined by certain boundary conditions ( e.g. in quantum dots or quantum billiards ) or by self - organization ( e.g. in atomic nuclei ) .
the shape of the nucleus is , indeed , an essential part of nuclear spectroscopic investigations and is characteristic of every nucleus . it may be spherical as well as oblate or prolate or even more complicated , see @xcite .
the spectroscopic properties of quantum systems are studied usually by means of a hermitian hamilton operator @xmath0 , the eigenstates of which are discrete .
due to the embedding of the system into a continuum of scattering wavefunctions , the lifetimes of its states become usually finite .
this agrees with the experimentally well - known finite lifetime of the states of a quantum system which is described in the standard hermitian quantum theory by means of the tunneling mechanism .
an exception are the well - known bound states in the continuum the lifetime of which is infinitely long .
these states occur under special conditions and are described well in different papers for different systems , e.g. @xcite .
moreover , the embedding of a quantum system into a continuum of scattering wavefunctions causes another effect which can , as a matter of principle , not be simulated in standard hermitian quantum theory : since every state of the system is coupled to the continuum , all the states of the system may mix via the environment .
this so - called external mixing ( em ) of the states is a second - order process and occurs additionally to the well - known direct ( internal ) mixing of the states which is contained in the hermitian hamiltonian @xmath0 .
experimentally , an example of em has been provided a few years ago in a mesoscopic system .
it has been shown in @xcite that two distinct quantum states are coupled through a common continuum . in a further experiment ,
the authors were able to show that em survives even under conditions of strongly nonequilibrium transport in the system @xcite .
a possible em of quantum states is very seldom considered in theory since it is fully beyond hermitian theory .
however , it can easily be taken into account when the properties of the system are described by a non - hermitian hamiltonian @xmath1 .
in such a case , em is involved explicitly in the eigenfunctions of @xmath1 , see @xcite .
in @xcite the , at first glance , unexpected result has been found that em needs not to be considered when the system is embedded in a common environment of scattering wavefunctions .
this result corresponds to the experience obtained from many numerical studies on realistic systems in which em is not at all considered .
also the influence of the singular exceptional points ( ep ) is not involved in these standard calculations . nevertheless , these results are not in contradiction with the statement of non - hermitian quantum physics that ep and em may influence , to a great extent , the spectroscopic properties of open quantum systems .
the point is that eps cause nonlinear processes in their neighborhood @xcite which are able to compensate the contributions that arise from the em of the states via the common environment @xcite .
correspondingly , the cross section shows the same resonance structure in the one - channel case when calculated with eps and em or without them .
the situation is however another one when the system is embedded not only in one common environment , but in several environments which exist independently of one another .
an example of such a situation is the decay of a nuclear state into different states of the residual nucleus by emission of a nucleon .
in such a case , we have the different widths @xmath2 ( called usually partial widths ) that characterize the decay of the state @xmath3 of the system in one of the open channels @xmath4 .
the different channels are defined by the decay of the state @xmath3 into different states of the residual nucleus by emission of a nucleon .
a channel is open when the energy difference between the state @xmath3 and the state of the residual nucleus is positive , i.e. when the emission of a particle is allowed .
another example is the transmission through , e.g. , a quantum dot . in this case , at least two different environments exist , namely that of the entrance channel and that of the exit channel .
several numerical studies performed with a non - hermitian hamilton operator @xmath1 on the basis of the tight - binding model have shown interesting non - trivial results @xcite : the transmission probability is anticorrelated with the phase rigidity of the eigenfunctions of the non - hermitian hamilton operator ( when averaged over energy in a certain energy window ) .
that means it depends on internal properties of the eigenfunctions of @xmath1 .
an enhancement of transmission through quantum dots or quantum billiards is found also in other studies , mostly by varying the coupling strength of the system to the environment . in @xcite ,
complex scaling is used while in @xcite the formation of a superradiant state under certain conditions is considered . neither in @xcite nor in @xcite , the enhancement of the transmission
is related to the existence of a special superradiant state ( that appears together with subradiant states ) .
the enhancement of the transmission in these studies is rather caused by the internal properties of the eigenstates of the non - hermitian hamilton operator @xmath1 , above all by the properties of its eigenfunctions .
it is the aim of the present paper to find typical spectroscopic features of an open quantum system that is coupled to two environments , which are independent of one another . in sect .
[ hamilt ] , we sketch the formalism used by us ; and in sect .
[ num ] we show some typical numerical results .
a discussion of the obtained results can be found in sect .
[ disc ] while some conclusions are drawn in sect .
[ concl ] . here , we bring into focus a possible experimental test of non - hermitian quantum physics .
to begin with , we sketch the features typical for an open quantum system embedded in one common continuum .
details can be found in @xcite and , above all , in @xcite .
they can be discussed by means of the @xmath6 non - hermitian matrix @xmath7 here , @xmath8 are two complex eigenvalues of the basic non - hermitian operator coupled to the environment @xmath9 ( called also channel @xmath9 ) @xcite .
the @xmath10 are the energies of the states @xmath3 and the @xmath11 are their widths .
the @xmath12 stand for the coupling matrix elements of the two states via the common environment @xmath9 .
they are complex where re(@xmath12 ) arises from the principal value integral and im(@xmath12 ) from the residuum @xcite .
the complex eigenvalues @xmath13 of @xmath5 give the energies @xmath14 and widths @xmath15 of the states of the localized part of the system . in ( [ ham2 ] ) ,
a singular point may appear , the so - called ep , at which two eigenvalues of @xmath5 coalesce @xcite . at these points ,
the two corresponding eigenfunctions are not orthogonal .
instead @xmath16 according to analytical and numerical results @xcite .
an ep is , according to its definition , related to the common environment in which the system is embedded . in other words
, it is well defined under the condition that the system is embedded in only one continuum .
the eigenfunctions of a non - hermitian hamilton operator are biorthogonal @xmath17 in the case of the symmetric @xmath6 hamiltonian ( [ ham2 ] ) , it is @xmath18 and the eigenfunctions should be normalized according to @xmath19 in order to smoothly describe the transition from a closed system with discrete states to a weakly open one with narrow resonance states . as a consequence of ( [ eif3 ] ) , the values of the standard expressions are changed , @xmath20 @xmath21 @xmath22 furthermore , the phase rigidity which is a quantitative measure for the biorthogonality of the eigenfunctions , @xmath23 while it approaches the value @xmath24 when an ep is approached . additionally to the hamiltonian ( [ ham2 ] )
, we will consider the non - hermitian matrix @xmath25 which describes the system without any mixing of its states via the environment .
in other words , @xmath26 corresponds to vanishing em of the eigenstates .
the eigenfunctions @xmath27 of @xmath5 can be represented in the set of eigenfunctions @xmath28 of @xmath29 , @xmath30 under the condition that the @xmath31 are normalized by @xmath32 .
the coefficients @xmath33 differ from the @xmath34 .
they contain the information on the strength of em .
the main features characteristic of open quantum systems are described well by the eigenvalues and eigenfunctions of ( [ ham2 ] ) .
typical values related to the eigenfunctions are the phase rigidity ( [ eif7 ] ) and the contribution of em ( [ eif12 ] ) to their purity .
all these values contain the influence of the environment .
they are proven experimentally , for details see @xcite .
the schrdinger equation @xmath35 may be rewritten into a schrdinger equation with source term @xcite , @xmath36 in this representation , the coupling @xmath37 of the states @xmath3 and @xmath38 of the localized system via the common environment of scattering wavefunctions ( em ) is contained in the source term of the schrdinger equation , for details see @xcite .
far from eps , the coupling of the localized system to the environment influences the spectroscopic properties of the system only marginally @xcite .
the influence is however nonvanishing also in this case , see e.g. the experimental results @xcite .
in the neighborhood of eps , the coupling between system and environment causes according to mathematical studies nonlinear effects in the schrdinger equation ( [ eif11 ] ) , see @xcite . among others ,
these effects lead to a conservation of the resonance structure of the cross section in the one - channel case which is unaffected by em and by the existence of eps .
thus , the one - channel case can not be used in order to test the results of the non - hermitian formalism .
let us consider the @xmath40 non - hermitian matrix @xmath41 here , @xmath42 and @xmath43 are the complex eigenvalues of the basic non - hermitian operator @xmath39 relative to channel @xmath9 and @xmath44 , respectively @xcite . usually @xmath45 .
the @xmath12 and @xmath46 stand for the coupling matrix elements of the two states via the environment @xmath9 and @xmath44 , respectively .
the eigenvalues @xmath47 and eigenfunctions @xmath48 of ( [ ham22 ] ) are characterized by two numbers : the number @xmath3 of the state ( @xmath49 ) of the localized part of the system and the number @xmath4 of the channel ( @xmath50 ) , called environment , in which the system is embedded .
usually , @xmath51 and @xmath52 . also the wave functions @xmath53 and @xmath54 differ from one another due to the em of the eigenstates via the environment @xmath9 and @xmath44 , respectively . that means , the system has four states , from a mathematical point of view .
the two environments are different from and orthogonal to one another .
further , the two eigenstates with equal @xmath3 and different @xmath4 arise from the same state @xmath3 of the localized part of the system .
the zeros in the matrix ( [ ham22 ] ) express the corresponding fact that the two eigenstates @xmath3 relative to the two channels 1 and 2 of @xmath39 can not interact with one another .
we mention here that the hamiltonian ( [ ham22 ] ) is formally the same as the hamiltonian ( 24 ) in @xcite .
there is however a fundamental difference : in ( 24 ) in @xcite , the whole system is _ fully _ embedded into both environments @xmath55 and @xmath56 which both are of completely different nature and exist independently of one another .
in contrast to this , the hamiltonian ( [ ham22 ] ) describes a system , the states of which are embedded partially in each of the two different environments .
these two different environments are also independent of one another . they are , however , nothing but _ parts _ of the total environment . without an ep in the considered parameter range in relation to both channels
, we have @xmath57 , @xmath58 and @xmath59 .
this case is realized in the decay of nuclear states since their decay probability can be determined only when the decaying state is well isolated from other states .
otherwise , new problems arise from the overlapping with other states @xcite . under the influence of an ep relative to @xmath60 and ( or ) relative to @xmath61
, the eigenvalues and eigenfunctions will be , however , different , @xmath51 , @xmath52 and @xmath62 in the corresponding parameter range .
such a situation may occur in transmission through a quantum dot or quantum billiard @xcite .
this case will be considered in detail in the following section .
in analogy to ( [ ham0 ] ) , we will consider also the non - hermitian hamiltonian @xmath63 which describes the system without any mixing of its states via any environment .
in other words , @xmath64 corresponds to vanishing em of the eigenstates via an environment .
the mixing of the eigenstates of ( [ ham22 ] ) can be represented in a set of eigenfunctions of ( [ ham00 ] ) in complete analogy to the relation ( [ eif12 ] ) for two states coupled to one common environment . using ( [ ham00 ] )
, we can write down the schrdinger equation with source term for the two - channel case in analogy to ( [ eif11 ] ) for the one - channel case .
the corresponding equation reads @xmath65 the source depends on the coupling of the system to both channels , i.e. on @xmath12 and on @xmath46 . under the special condition @xmath66 , the source term of ( [ eif11s4 ] ) depends on the only value @xmath67 .
hence it is similar to that of the one - channel case ( [ eif11 ] ) .
the situation is completely different in the general case with two channels ( two environments ) in which @xmath68 .
we will show results of numerical calculations in the following section .
we repeat here that , according to their definition @xcite , eps occur only in the one - channel case , i.e. only in the two submatrices related to channel @xmath9 and channel @xmath44 , respectively .
they are not defined in the @xmath40 matrix ( [ ham22 ] ) .
however , the eps of the two submatrices in ( [ ham22 ] ) may influence the dynamics of the open two - channel system .
the resonance structure of the @xmath69 matrix containing two resonance states , can be obtained from the expression @xmath70 which is derived in @xcite for the case that two resonance states are coupled to a common continuum of scattering wavefunctions , see also @xcite . the expression ( [ sm4 ] )
is unitary .
according to ( [ sm4 ] ) , the resonance structure of the cross section is determined exclusively by the spectroscopic values of the localized part of the system , i.e. by the eigenvalues @xmath71 of the non - hermitian hamiltonian @xmath1 .
the expression ( [ sm4 ] ) allows us therefore to receive reliable results also in the two - channel case and , moreover , when the phase rigidity of the eigenfunctions of @xmath1 is reduced , @xmath72 , and when the eigenfunctions of @xmath1 contain em , i.e. when they are mixed in the set of eigenfunctions @xmath28 of @xmath73 according to ( [ eif12 ] ) .
as shown in @xcite , the @xmath69 matrix contains , generally , nonlinear effects which are caused by an em of the resonance states via the environment .
the one - channel case does , however , not allow us to prove the existence of these nonlinear effects and of em , since the resonance structure of the cross section calculated with and without em is the same in this case @xcite .
this result agrees , on the one hand , with the experience received from many different numerical studies in realistic cases which are performed without taking into account em . on the other hand , it is not in contradiction with the conclusions received from the study of non - hermitian physics of open quantum systems . according to the results obtained in @xcite , the evolution of the system near to an ep is driven exclusively by the nonlinear source term of the schrdinger equation ( [ eif11 ] ) which describes the coupling of the localized part of the system to the common environment and is characteristic of the open quantum system embedded in one environment .
the calculations in @xcite are performed without varying @xmath37 , i.e. @xmath37 can not be responsible for the width bifurcation occurring in these calculations under the influence of an ep . obviously , the nonlinear source term is the driving force .
it is able , in the one - channel case , to largely conserve the resonance structure of the cross section .
the conservation of the resonance structure of the cross section which is possible in the one - channel case , is expected to be impossible , generally , in the two - channel ( or more - channel ) case according to ( [ eif11s4 ] ) since , generally , @xmath74 .
this gives the chance to test non - hermitian quantum physics by means of the two - channel case .
we will provide numerical results and discuss their physical meaning in the following sections of the present paper .
for illustration , we show in fig . [ fig1 ] a few typical numerical results for two states in the two - channel case .
the results of the left column are obtained for the special case @xmath75 and those of the right column for the general case @xmath76 .
the numerical results for the eigenvalue trajectories @xmath77 and the eigenfunction trajectories @xmath78 of the hamiltonian @xmath39 are obtained by starting from parameter - dependent energies @xmath79 and parameter - independent widths @xmath80 .
we consider two different cases : in one case , the widths @xmath80 of the two states are equal , @xmath81 ( correspondingly to @xmath75 ) , while they are different from one another in the other case , @xmath82 . the eigenvalue trajectories figs .
[ fig1].a , b and [ fig1].f , g , respectively , show two eps between which the widths @xmath83 bifurcate . at the critical parameter value @xmath84 width bifurcation is maximum . here
the phase rigidity approaches the value @xmath9 ( figs .
[ fig1].d and [ fig1].i ) , meaning that two states become ( almost ) orthogonal at this parameter value .
the em of the states via the continuum can not be neglected at this parameter value ( figs .
[ fig1].e and [ fig1].j ) .
while the eigenvalue trajectories of @xmath39 are influenced by an ep mainly in its very neighborhood , the eigenfunction trajectories are influenced in a comparably large parameter range .
this can be seen , above all , in the phase rigidity ( figs .
[ fig1].c , d , h , i ) which is a quantitative measure for the biorthogonality of the eigenfunctions of @xmath39 .
this holds true in the one - channel case @xcite as well as in the two - channel case considered in fig .
[ fig1 ] .
although an ep is defined in relation to only one channel , a hint to its position can be seen in its neighborhood in the eigenfunctions of @xmath39 : the phase rigidity @xmath85 decreases , indeed , at the position of an ep .
it does however not approach the value zero ( figs .
[ fig1].c and [ fig1].h ) .
this is true not only when @xmath81 but also when @xmath86 .
similar results are obtained by starting from parameter - independent energies @xmath79 and parameter - dependent widths @xmath80 . in difference to the results shown in fig .
[ fig1 ] , the eigenvalues show , in this case , level repulsion ( appearing in the @xmath87 ) between the two eps , while the widths @xmath83 of the two states are the same in this parameter range .
these results correspond to those obtained in the one - channel case @xcite .
interesting information on the spectroscopic properties of the localized part of the system is contained in the resonance structure of the cross section which can be observed experimentally .
we consider the transition from channel @xmath9 to channel @xmath44 which simulates transmission . using ( [ sm4 ] ) , we have calculated the resonance structure of the transmission at the critical parameter value @xmath88 and at two values of @xmath89 beyond the critical range . in fig .
[ fig2 ] , the results without em ( corresponding to @xmath90 ) are shown .
there appear four states , indeed , according to the results of analytical studies .
characteristic of the resonance structure of the cross section of a two - level system coupled to _
one _ channel is its double - hump structure around maximum width bifurcation @xcite .
[ fig3 ] and [ fig4 ] show the transmission through a two - level system coupled to _ two _ channels in the analog parameter range around maximum width bifurcation .
these figures illustrate that the double - hump structure of the cross section is restored in the transmission ( i.e. in the two - channel case ) under the influence of em . in fig .
[ fig3 ] ( with different coupling strengths @xmath91 according to the eigenvalues and eigenfunctions of fig .
[ fig1 ] right ) , the double - hump structure around the maximum width distribution is very similar , indeed , to that of the cross section in the one - channel case ( solid black line in fig .
[ fig2 ] compared to solid black line in fig .
4.a in @xcite ) .
the contour plot of the transmission is however much richer than the cross section in the one - channel case .
this is reflected also in the dotted ( blue ) and dashed ( red ) lines of fig .
[ fig3].a .
more interesting are the results shown in fig .
[ fig4 ] corresponding to the eigenvalues and eigenfunctions of fig .
[ fig1 ] left with @xmath92 . here
, the double - hump structure also appears .
the bumps are however broadened in energy . due to this broadening ,
the transmission is enhanced in a comparably large energy window . in this energy window ,
the phase rigidity averaged over energy , is reduced ( see fig .
[ fig1 ] ) .
the results shown in figs . [ fig1 ] to [ fig4 ] show very clearly that the mixing of the states via the environment ( em ) plays an important role in open quantum systems .
this mixing is a second - order effect and does not appear in standard hermitian quantum physics . while its influence onto the spectroscopic properties of the localized part of the system far from singular points is usually small , it becomes important and determines the dynamics of the system in a finite neighborhood of singular points .
such points are the well - known exceptional points ( eps ) . according to mathematical studies @xcite ,
eps are defined in relation to _
one _ continuum .
their physical meaning is studied therefore , up to now , mostly in systems that are embedded in one well - defined common environment .
the results of many different papers show their meaning for the dynamical behavior of open quantum systems , among others for phase transitions . for references
see the review @xcite .
most states of physical systems are coupled , however , to different channels ( environments ) .
for example , the transmission through a small system ( e.g. a quantum dot ) is related to , at least , two channels , the entrance and the exit channel .
the transmission through such systems is studied by many authors in many different papers by using different methods . in the present paper , it is studied for the first time in the framework of the non - hermitian quantum physics without additional assumptions ( such as , e.g. , tight - binding ) by taking into account that the ( localized ) system is embedded in more than one environment , i.e. that it is coupled to more than one channel .
the results are compared to those known for open quantum systems embedded in one common environment .
the similarities and differences are the following . 1 .
far from singular points , the spectroscopic properties of the system are influenced only marginal by its embedding into an environment .
this result is independent of the number of environments .
2 . in the neighborhood of singular points ( eps )
, the dynamical behavior of the system is determined by nonlinear processes which are caused by the existence of the singular points .
the nonlinear processes restore , in any case , the resonance structure of the cross section obtained without em .
this causes the possibility to describe the system without taking into account em and eps in the _ one - channel _ case @xcite . in the _ two - channel _ case
, however , it gives rise to an enhancement of the transmission at maximum width bifurcation ( or maximum level repulsion ) .
this effect is measurable .
an experimental test of the formalism of the non - hermitian description of open quantum systems is therefore possible in the two - channel ( or more - channel ) case , while it is impossible @xcite in the one - channel case .
although em is a second - order effect , it may become important in the neighborhood of eps due the nonlinear processes dominating the dynamics in this parameter range .
we underline once more that these processes are related to internal properties of the eigenstates of a non - hermitian hamilton operator .
they are _ not _ involved in any version of hermitian quantum physics . according to early studies @xcite of open quantum systems by means of a non - hermitian hamilton operator ,
short - lived states appear together with long - lived ones at high level density even without changing the coupling strength between system and environment .
these states result from the phenomenon resonance trapping occurring in the regime of overlapping resonances @xcite . in some papers
, this phenomenon is studied as a function of the coupling strength between system and environment and is called segregation of decay widths @xcite . here
the short - living states are called superradiant states and the long - lived ones subradiant states .
this formalism is applied also to the description of transmission , see e.g. @xcite . in this formalism
, an enhancement of transmission is related to the existence of a superradiant state ( together with subradiant states ) .
in contrast to this interpretation , the enhancement of transmission in a certain critical parameter range is , according to the results obtained by us in the present paper , _ not _ caused by the existence of a superradiant state . correspondingly , subradiant states do not appear in our calculations .
instead , the enhancement of the transmission in a certain energy window occurs , in our calculations , due to nonlinear processes .
it is related to the non - rigidity of the phases of the eigenfunctions of @xmath93 in the neighborhood of eps and to the em of the states via the environment .
the numerical results show clearly that an enhancement of the transmission in a certain energy window is correlated with a reduction of the phase rigidity , when averaged over energy in this window .
this result is the same as that obtained in @xcite . in this paper , the transmission through a system with many states
is calculated numerically by using the tight - binding lattice model ( according to datta @xcite ) with one channel in each of the two attached identical leads . as a result of this calculation , the phase rigidity , averaged over the considered energy window , and the transmission
are anti - correlated .
thus , the two very different methods ( each of which solves the problem exactly ) produce qualitatively the same results .
considering the results of the present paper , we arrive at the following conclusion .
a direct experimental test of the formalism of non - hermitian quantum physics is possible by investigating the question whether or not the observed enhanced transmission in a certain parameter range is related to the existence of a superradiant state .
according to theory @xcite , a superradiant state appears if the coupling strength between system and environment increases .
it appears always together with subradiant long - lived states .
in contrast to this statement , our results show that an enhanced transmission in a certain parameter region is caused by the properties of the eigenfunctions of the non - hermitian hamilton operator @xmath1 , above all by the non - rigid phases of its eigenfunctions .
a special state which is strongly coupled to the environment and is correlated with a partial decoupling of other states from the environment , does not exist .
summarizing we state that open quantum systems embedded in two ( or more ) environments are abounded in very many interesting aspects of non - hermitian quantum physics .
they allow , on the one hand , to test experimentally the basic features of non - hermitian quantum physics . on the other hand
, they provide the possibility of many promising applications .
in contrast to the definition that is used in , for example , nuclear physics , we define the complex energies before and after diagonalization of @xmath93 by @xmath94 and @xmath95 , respectively , with @xmath96 and @xmath97 for decaying states . this definition will be useful when discussing systems with _
( positive widths ) and _ loss _ ( negative widths ) . | calculations for open quantum systems are performed usually by taking into account their embedding into one common environment , which is mostly the common continuum of scattering wavefunctions .
realistic quantum systems are coupled however often to more than one continuum .
for example , the conductance of an open cavity needs at least two environments , namely the input and the output channel . in the present paper ,
we study generic features of an open quantum system coupled to two channels .
we compare the results with those characteristic of a one - channel system . of special interest
is the parameter range which is influenced by singular ( exceptional ) points . here , the states of the system are mixed via the environment and the dynamics is determined , in any case , by nonlinear processes .
these processes restore the resonance structure of the cross section in the one - channel case , such that there is no need to consider them in calculations for concrete systems . in the two - channel case , however , they cause new effects .
most interesting is the enhanced conductance of an open cavity in this parameter range .
this enhancement follows directly from the properties of the eigenstates of a non - hermitian hamilton operator ( and is not related to the existence of a superradiant state ) .
the study of the conductance allows , among others , an experimental test of the non - hermitian quantum physics . |
with the data from cern s lep electron - positron collider , desy s hera electron - proton collider and fermilab s tevatron proton - antiproton collider , the standard model ( sm ) has received overwhelming experimental support .
no new physics has been observed essentially at mass below @xmath3 .
thus , it is imperative that the fermilab tevatron , the world s highest energy particle collider , be used for searches of new phenomena .
the cdf and dexperiments have been actively studying their data for evidence of previously unobserved particles .
presented below are the latest results of searches at the tevatron for supersymmetric particles ( chargino , gluino ) , leptoquark bosons and heavy gauge bosons .
the minimal supersymmetric standard model ( mssm ) @xcite is a _ supersymmetrized _ sm with two higgs doublets , which is one of the most appealing theories proposed to test grand unification @xcite .
conservation of @xmath6-parity requires the susy particles to be produced in pairs and prevents decays of the lightest supersymmetric particle ( lightest neutralino @xmath7 ) .
the most distinctive signatures of chargino - neutralino ( @xmath4 ) and @xmath5 ( @xmath8 , @xmath9 ) pair production are trilepton events @xcite and like - sign ( ls ) dileptons associated with large missing transverse energy ( @xmath10 plus multi - jets @xcite .
pair - produced @xmath4 decay as @xmath11 and @xmath12 .
the striking signature of these events is thus three isolated leptons plus @xmath13 .
the cdf search is based on 100 data .
most of the events are from inclusive @xmath14 and @xmath15 triggers at @xmath16 10 .
the trilepton requirements are @xmath17 11 and @xmath18 5 ( 4 ) for @xmath14 ( @xmath15 ) , which is the same as in the previous analysis ( based on 19 ) @xcite .
eight events are left which are consistent with the sm background .
a further cut of @xmath19 15 gev is imposed to reduce the background .
there is no trilepton event candidate found , which is consistent with an expected background of @xmath20 for four trilepton ( @xmath21 , @xmath22 , @xmath23 , @xmath24 ) modes .
the d trilepton analysis requires a lower cut of @xmath25 5 , although the events are from several trigger paths ( inclusive @xmath14 and @xmath15 samples and @xmath26 , @xmath27 and @xmath28 dilepton samples ) .
the acceptance difference due to those trigger thresholds are corrected using a monte carlo simulation .
dalso observes no trilepton event candidate in 14 , which complies with an expected background of @xmath29 for @xmath21 , @xmath30 for @xmath22 , @xmath31 for @xmath23 , @xmath32 for @xmath24 @xcite .
both cdf and dhave a similar constraint inspired by supergravity models @xcite . the cdf analysis @xcite calculates slepton ( @xmath33 ) and sneutrino ( @xmath34 ) masses from @xmath35 , @xmath36 , and @xmath37 using the renormalization group equations @xcite , while d@xcite uses a minimal supergravity model in isajet @xcite . in those models , chargino and neutralino have three - body decays .
the cdf analysis excludes @xmath38 66 ( 95% c.l . ) at @xmath35 = 2 , @xmath39 and @xmath40 gev ( the region of maximum experimental sensitivity ) .
the cdf limit on @xmath41 ( 66 ) is comparable to the lep1.5 result @xcite . the d result is not sensitive to set any mass limits for the isajet model .
limits on @xmath42(4 trilepton modes ) are also obtained : 0.6 pb ( cdf ) and 4 pb ( d ) for 70-chargino .
cdf also examines one particular supergravity model , flipped su(5 ) model @xcite . in this model ,
@xmath43 , so that the trilepton signal is nearly maximized via @xmath44 66% ( @xmath14 and @xmath15 ) and @xmath45 = 100% .
however , the mass difference of @xmath46 and @xmath7 decreases as @xmath47 increases , so that the total trilepton acceptance as a function of @xmath47 becomes flat at about 5% at 60 and falls off for @xmath48 75 .
figure [ upp_lim ] shows the 95% c.l .
upper limit curve on @xmath49 .
the points in the figure are the isajet predictions for @xmath50 production in the supergravity model .
the limit on @xmath47 is 73 .
the ls dilepton approach is complementary to the classic multi - jets+@xmath13 analysis in the search for @xmath51 , @xmath9 and @xmath52 production .
the signature arises mainly from @xmath5 production followed by dominant decays of @xmath53 and @xmath54 if @xmath55 .
thus , the final state contains two or more leptons ( from @xmath41 and @xmath56 decays ) , @xmath13 and jets . since the gluino is a majorana fermion ,
there is no charge correlation between these leptons .
such a data analysis begins with a dilepton sample ( @xmath57 12 ( 10 ) ) and subsequently requires @xmath58 25 gev and @xmath59 with @xmath60 15 gev in @xmath61 since the production cross section times branching ratio is small , cdf searched for the dilepton signature without the ls requirement . in 19 , one event ( @xmath62 ) is observed , which is consistent with an expected background of @xmath63 . the lower gluino mass at 95% c.l . is calculated to be 154 for @xmath64 and 224 for @xmath65 at @xmath66 and @xmath40 gev @xcite .
those limits weakly depend on the @xmath35 values ( 2 - 8 ) and the @xmath15 values ( 200 gev @xmath67 1000 gev ) .
leptoquark is a generic term for color - triplet bosons which couple both to a quark and a lepton .
they appear in many sm extensions which join the quark and lepton sectors at more fundamental levels @xcite . both cdf and dsearch for pair - produced lepton quarks .
the signatures are @xmath68 + 2 jets and @xmath69 + @xmath13 + 2 jets .
the danalyses of the first and second generation scalar leptoquarks ( @xmath70 and @xmath71 ) searches using @xmath72 and @xmath73 events are based on 15 data @xcite .
neither analysis finds evidence for a leptoquark signal .
the mass limits are 130 ( 116 ) at @xmath74 for @xmath70 with hmrs - b ; 119 ( 97 ) at @xmath74 for @xmath71 with cteq2pm . here
= @xmath76 . note that mt - lo is the nominal choice in d s @xmath70 and @xmath71 analyses .
the cdf results on the @xmath70 and @xmath71 searches are based on 4.1 @xcite and 67 data @xcite .
the limits at @xmath74 are 113 ( 80 ) for @xmath70 with hmrs - b and 180 ( 141 ) for @xmath71 with cteq2pm .
cdf also searches for the third generation leptoquark ( @xmath77 ) in @xmath78 mode , where one of the @xmath79 leptons decays semileptonically and the second @xmath79 has 1 or 3-prong hadronic decays .
the @xmath79-decay lepton ( @xmath14 or @xmath15 ) is required to have @xmath80 20 ( @xmath81 ) .
the hadronic decay @xmath79 must have uncorrected @xmath82 15 gev ( @xmath83 ) associated 1 or 3 tracks ( the leading track @xmath80 10 ) in a @xmath84 cone around the energy cluster .
the charge of the @xmath79 candidate is defined to be the total charge of 1 or 3 tracks and required to be an opposite sign to the lepton charge .
this reduces the qcd background by a factor of @xmath852 .
the azimuthal separation between the lepton and @xmath13 directions should be less than @xmath86 to reduce the @xmath87 + jets events .
finally , two jets are required with uncorrected @xmath82 10 gev in @xmath88 .
no events are left ; consistent with the expected sm background of @xmath89 events including 1.0 @xmath90 event .
figure [ fig : cdf_lq3 ] shows the results of the mass limits on @xmath77 for the scalar boson and the vector bosons ( both @xmath91 = 0 and 1 , where @xmath91 is an anomolous chromomagnetic moment ) .
all limits on @xmath92 are obtained at @xmath75 = 1 .
in extended gauge models @xcite proposed to restore left - right symmetry to the weak force , the right - handed , charged bosons @xmath94 can decay with large probability to right - handed @xmath95 pairs .
both dand cdf have searched for the heavy @xmath94 through the process @xmath96 ( or @xmath97 if @xmath98 in 74.4 and 19.7 , respectively .
the searches were made by assuming a standard strength of the coupling and the decay @xmath99 to be suppressed by a left - right mixing angle @xmath100 = @xmath101 [ @xmath102@xmath103 where @xmath101 is @xmath104(1 ) .
the limits are 720 @xcite and 652 @xcite
. the cdf experiment is conducting a complementary search for @xmath94 in the decay to @xmath105 with @xmath106 30 ( @xmath107 1.05 ) , @xmath19 30 gev , @xmath60 50 ( 20 ) gev .
additional tracking , isolation and electron identification criteria are also imposed .
the previous result @xcite is updated using 110 data : 512 @xmath94 candidate events are left .
we observe 7 events for @xmath108 600 , compared to 4.2 expected events .
we find no significant evidence for excess @xmath109 production produced in association with a @xmath110 .
the limits are obtained by a likelihood fit with the data , background and signal shapes .
the results of the excluded region in the @xmath100-@xmath111 plane are shown in fig .
[ fig : cdf_wprime_wz ] .
the range of 200 - 560 is excluded at @xmath100 = 1 , which is consistent with the limits above . however , the cdf analysis is insensitive to setting any limit at @xmath101 = 1 .
neutral gauge bosons in addition to the @xmath113 are expected in many extensions of the sm @xcite .
these models typically specify the strength of the couplings of such bosons to quarks and leptons but make no mass prediction .
@xmath114 bosons may be observed directly via their decay to lepton pairs ( @xmath115 , @xmath116 ) .
the cdf and dresults presented below are derived assuming sm coupling strengths . for @xmath117 ,
the cdf criteria are @xmath106 25 gev in @xmath118 .
the d requirements are similar : @xmath106 30 gev in @xmath119 .
additonal tracking , electron identification and isolation cuts are imposed . the only appreciable background is from drell - yan @xmath120 in the high mass region . for @xmath121 250 , cdf ( d ) observes 8 ( 1 ) events in 67.6 ( 14.4 ) data which is consistent with the expected background of 8.1 ( 1.4 ) events . for @xmath122 ,
the cdf criteria are @xmath123 20 gev in @xmath124 with additional quality cuts . in 67 data , seven events with @xmath125 200
are observed consistent with the expected rate from drell - yan production .
the mass limits are extracted by a binned maximum likelihood analysis on the data comparing the data to a sum of the drell - yan background and the @xmath114 expectation .
the drell - yan and @xmath114 distributions are modeled by the leading order monte carlo .
the cdf mass limits on @xmath114 are 620 for @xmath126 and 590 for @xmath127 . at this time , a combined limit of 650 is available from 67.6 @xmath128 data and 71.3 @xmath127 data @xcite .
the dmass limit from @xmath129 data is 490 @xcite .
the cdf and dexperiments have searched for various new physics phenomena . in these studies , there is no evidence for physics beyond the standard model .
p. langacker , proceedings of pascos 90 symposium , eds .
p.nath and s.reucroft ( world scientific , singapore , 1990 ) ; j. ellis , s. kelley , and d.v .
nanopoulos , phys . lett . *
b249 * , 441 ( 1990 ) ; u. amaldi , w. de boer , and h. f@xmath130rstenau , phys .
b260 * , 447 ( 1991 ) ; p. langacker and m .- x .
luo , phys . rev . *
d44 * , 817 ( 1991 ) . p.
nath and r. arnowitt , mod .
a2 * , 331 ( 1987 ) ; r. barbieri , f. caravaglios , m. frigeni and m. mangano , nucl .
b367 * , 28 ( 1991 ) ; h. baer and x. tata , phys . rev . * d47 * , 2739 ( 1993 ) ; j.l .
lopez , d.v .
nanopoulos , x. wang and a. zichichi , phys . rev . *
d48 * , 2062 ( 1993 ) ; h. baer , c. kao , and x. tata , phys . rev . *
d48 * , 5175 ( 1993 ) . for a recent review ,
see r. arnowitt and p. nath , `` supersymmetry and supergravity : phenomenology and grand unification , '' proceedings of vii@xmath131 j.a .
swieca summer school , campos de jordao , brazil , 1993 ( world scientific , singapore , 1994 ) .
iba@xmath132ez , c. lopez , and c. muo@xmath132z , nucl .
* b256 * , 218 ( 1985 ) .
ross and r.g .
roberts , nucl .
phys . * b377 * , 571 ( 1992 ) ; r. arnowitt and p. nath , phys .
lett . * 69 * , 725 ( 1992 ) ; s. kelley , j.l .
lopez , d.v .
nanopoulos , h. pois , and k. yuan , nucl . phys . *
b398 * , 31 ( 1993 ) ; g.l .
kane , c. kolda , l. roszkowski , and j.d .
wells , phys . rev . * d49 * , 6173 ( 1994 ) .
we used an approximate formula ( @xmath33 and @xmath34 masses to @xmath133 and @xmath134 masses ) in h.baer _ et al .
_ , phys . rev . *
d47 * , 2739 ( 1993 ) .
h. baer , f.e .
paige , s. d. protopopescu and x. tata , _ proceedings of workshop on physics at current accelerators and the supercollider _
, eds . j. hewett , a. white and d. zeppenfeld ( argonne national laboratory , 1993 ) .
dcollaboration , phys .
72 * , 965 ( 1994 ) .
the dtotal integrated luminosity has changed since the publication due to an update of the total inelastic cross section ( averaged new cdf and e710 ) used to calculate the luminosity .
the new luminosity is 13.8 . | various recent results of new particle searches at the fermilab tevatron are presented .
no evidence is found for supersymmetric particles ( chargino , gluino ) , leptoquark bosons and heavy gauge bosons in @xmath0 collisions at @xmath1 = 1.8 tev .
excluded mass regions for each particle are determined .
u l # 1 # 1#2 -0.6 cm
-0.2 cm -0.2 cm @xmath2 teruki kamon + department of physics , texas a&m university + college station , texas 77843 - 4242 + on behalf of the cdf and dcollaborations + |
the decimetric synchrotron emission from galaxies is believed to arise from relativistic electrons accelerated by shock waves either in the interstellar medium , or in structures associated with an active galactic nucleus ( agn ) . in the former case
, there is evidence that the decimetric luminosity is proportional to the rate of massive star formation , the link arising because supernovas are perhaps the dominant source of accelerating shock waves @xcite . in the later case , the decimetric luminosity of an agn may be related to the physical properties of the underlying black hole , and the rate and mode of mass accretion by that black hole @xcite .
observations of the total decimetric luminosity of a galaxy thus allow an estimate of the extinction - free star - formation rate in galaxies where agn emission can be considered negligible .
conversely , where the star formation rate is relatively low they provide information on the character of activity excited by a nuclear black hole .
a few galaxies emit significant decimetric from combined nuclear star - formation and agn activity @xcite .
the knowledge provided by radio flux density measurements is greatly enhanced when optical spectroscopic data are also available .
optical spectra reveal the galactic redshift , allowing the calculation of luminosities and hence estimates of the cosmic evolution of the rates of star - formation and black hole mass accretion .
they also reveal the age and other properties of the most optically luminous components of the galactic stellar population and , in some cases , the state of excitement of a substantial fraction of the interstellar medium of the galaxy .
investigations of large galaxy samples having combined optical and radio data offer opportunities to address several outstanding astrophysical problems .
the 2df galaxy redshift survey ( 2dfgrs , * ? ? ?
* ) has measured the optical spectra of over 220,000 galaxies photometically selected to have an extinction - corrected magnitude brighter than @xmath5=19.45 .
the survey comprises strips at the equator , and near the south galactic pole ( sgp ) .
the sgp strip densely covers @xmath6 ( @xmath7 , @xmath8 ) , along with 99 randomly placed @xmath9 ( diameter ) fields .
@xcite studied the relatively bright radio sources among the 2dfrgs objects by cross - matching the 1.4 ghz nrao vla sky survey ( nvss , * ? ? ? * ) with the first @xmath4 20% of the 2dfgrs .
approximately 1.5% of the 2dfgrs galaxies ( @xmath4 5% nvss sources ) have radio counterparts at the nvss radio flux density detection limit of 2.8 mjy .
presumably , the majority of unmatched ( 95% ) nvss 1.4 ghz sources are agns whose optical counterparts are too faint to be detected at @xmath5=19.45 ( c.f . * ? ? ?
some 60% of the identifications are agns ( radio galaxies and some seyferts ) and the remainder 40% comprised of star - forming objects .
@xcite note that complete cross - matching of the 2dfgrs and nvss will provide approximately 4,000 radio source spectra , a sample large enough to measure radio galaxy evolution to @xmath10 and to locate the most luminous star - forming galaxies to @xmath11 .
@xcite have reported evidence at the level of two standard deviations for evolution in the rate of star formation density between the mean redshifts , @xmath12 0.02 and 0.06 , of their nvss / ugc sample and the nvss/2dfgrs sample of @xcite , respectively .
there is considerable interest in seeking radio identifications of 2dfrgs objects below the nvss flux density limit .
spectroscopy of the optical counterparts of radio sources found in sub - mjy surveys ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) suggests that the faint radio population will be a mixture of star - forming galaxies and agns , as it is for the mjy radio identifications .
interestingly , 1.4 ghz radio source counts rise sharply towards the euclidean rate near 1 mjy ( e.g. * ? ? ?
* ) , a phenomenon ascribed to the rising proportion of star - forming galaxies in the sub - mjy population @xcite . in view of the tight correlation between radio and far infra - red ( fir )
luminosity among the star - forming population , these galaxies likely correspond to the population of lower star formation rate `` iras galaxies '' seen at apparently brighter magnitudes
. studies of the sub - mjy counterparts of the 2dfgrs objects will thus encompass star - forming objects down to @xmath13 whz@xmath14 out to redshifts of @xmath15 , where evolution appears to have occurred , as well as agns of moderate radio luminosity .
the sgp part of 2dfgrs covers over 900 square degrees and would require a very large allocation of radio telescope time to undertake a co - extensive sub - mjy survey . acknowledging this , we have undertaken a pilot study of a selected region of the 2dfgrs ( equivalent area of one 2dfrgs - field ) , employing the australian telescope compact array ( atca ) at 1.4 ghz to explore probable outcomes and optimal observing strategies for a more extensive survey .
section @xmath16 describes the survey field selection , observations and data reduction and @xmath17 the procedures used to derive radio flux measurements and radio upper limits of the 2dfgrs objects .
@xmath18 presents the radio luminosity function derived from our sample and comparison with that of @xcite . in @xmath19
we explore the radio detection rate of 2dfgrs sources as a function of survey depth and in @xmath20 we consider the optimisation of a full - scale sub - mjy survey . throughout the paper we have used @xmath21 , @xmath22kms@xmath14mpc@xmath14 , and adopted for all sources , a radio spectral index @xmath23 ( @xmath24 ) .
one aim of this pilot survey is to investigate a survey strategy which yields optimal benefit from the atca observing time that would be invested in a deep , large - scale radio follow - up of the 2dfgrs .
the basic trade - off lies between the steeply rising surface density of sources at fainter survey limits , versus the quadratic dependence of survey sensitivity on integration time .
this implies a strategy in which a small area is surveyed very deeply , and a progressively larger area is surveyed to a progressively poorer sensitivity . given the breadth of the total radio luminosity functions for both star - forming and agn galaxies , and the weak dependence of the bi - variate radio luminosity function on optical luminosity , it would be hard to argue that any flux density range is to be preferred over any other , suggesting a strategy in which equal numbers of sources are detected to a progressively changing flux density limit .
table 1 presents the results of calculations designed to illustrate this strategy .
the table uses the published sensitivity of the atca at 1.4 ghz using two 128 mhz if channels , namely that a point source of @xmath25 mjy is detected at the field centre with @xmath26 sensitivity in a time of @xmath27 minutes , provided that full @xmath28 coverage is obtained ( atca on - line sensitivity calculator ) .
the table relies on integral source densities read from fig .
2(b ) in @xcite .
according to the table , a single pointing observed for 5,000 m ( 83 h ) would reveal just over 3,000 sources per square degree , brighter than 30 @xmath1jy , and approximately the same number of sources would be detected to a @xmath26 limit of 1 mjy if an area of 34 pointings was surveyed for a total of 170 m ( 3 h ) .
these values are ideal estimate , excluding observation overheads . in practice , there are several other factors that must be considered in designing the survey .
the limits given in the table refer to the most sensitive part of the primary beam , and can be approached to within a factor of about @xmath29 over a large area only if mosaicing techniques are exploited .
observations must be staged to provide adequate @xmath28 coverage , and bright radio sources must be avoided to achieve high dynamic range . | the australia telescope compact array ( atca ) has been used to survey at 1.4 ghz , a small region ( @xmath0 3 sq deg ) overlapping with the 2df galaxy redshift survey @xcite .
we surveyed with a varying radio sensitivity , ranging from 1 mjy 20 @xmath1jy ( @xmath2 ) .
there are 365 2dfgrs sources with z @xmath3 0.001 lying within the surveyed region , of which 316 have reliable spectral classification .
following @xcite , we visually classified 176 as agn or early - type galaxies , and 140 as star - forming galaxies .
we derived radio flux density measurement or upperlimits for each of the 365 2dfgrs sources .
the fraction of radio detected 2dfgrs star - forming galaxies increases from @xmath4 50% at @xmath4 0.7 mjy up to @xmath4 60% at @xmath4 0.2 mjy .
the mean redshift for the fraction of radio detected star - forming galaxies increases with increasing radio detection sensitivity , while the mean redshift is fairly constant for the agn / early - type fraction .
we found very similar radio detection rates of 2dfgrs galaxies for both the agn / early - type and star - forming components .
the radio detection rate increases approximately linearly with respect to the rate of increase in radio detection sensitivity .
we derived the radio luminosity function for our sample and it was found to be consistent with that of @xcite .
we have also compared the total flux densities of nvss sources common to our survey , and we discuss strategies for a large - scale radio survey of the 2dfgrs sample .
surveys galaxies : statistics galaxies : evolution galaxies : distances and redshifts galaxies : fundamental parameters
radio continuum : galaxies |
we consider stochastic simulation ( or monte carlo simulation ) of the solution of a dirichlet boundary value problem of a time - independent schrdinger equation with constant potential . for positive potentials this equation
is called the yukawa equation or the linearized poisson boltzmann equation . for zero potential
the equation is known as the laplace equation . for negative potential
it is known as the helmholtz equation .
the connection between the dirichlet boundary value problems and the brownian motion date back to kakutani @xcite , who provided a stochastic representation of the solution of the laplace equation with dirichlet boundary conditions in terms of the exit locations of a brownian motion .
later , this connection has been extended for the schrdinger equation , cf . @xcite and references therein .
these stochastic representations for constant non - zero potentials include the exit time of the brownian motion in addition to the exit location ; and in the case of non - constant potential , the stochastic representation depends on the entire path of the brownian motion up to the time it exits the domain .
stochastic representations provide monte carlo simulation methods for the solutions of the boundary value problems .
these methods are especially attractive in high dimensions where deterministic methods are typically extremely costly .
the obvious idea to use the stochastic representations is to simulate brownian particles on the domain with a fine time - mesh .
this is , however , very costly computationally . in order to avoid the simulation of the trajectory of the brownian particles very precisely ,
different algorithms have been proposed . in this paper
, we consider different walk on spheres algorithms that simulate the brownian motion only on successive spheres in the domain .
if one only needs to simulate the exit location , and not the exit time , of the brownian motion , one can use the very efficient classical walk on spheres ( wos ) algorithm due to muller @xcite .
unfortunately , the stochastic representation involving only the exit locations of brownian motion corresponds precisely the laplace equation .
this is our motivation to transform the constant - potential schrdinger equation into a laplace equation : to make the classical wos algorithm applicable . also , this transformation should be of interest in its own right . the transformation , the so - called duffin correspondence , removes the constant potential in the schrdinger equation with the cost of adding one extra dimension to the boundary value problem .
the idea of the correspondence is due to duffin @xcite , where the correspondence was used for the yukawa equation , i.e. for the case of positive constant potential , on the plane .
this was later extended to general euclidean spaces in @xcite . in this paper , we extend the duffin correspondence to cover also the helmholtz case , i.e. negative constant potential , in general euclidean spaces .
let us note that there are already efficient modified wos algorithms that simulate the exit time ( or its laplace transform ) and the exit location of the brownian motion that can be used for the constant - potential schrdinger equation studied here .
indeed , such stochastic simulation algorithms have been studied excessively ; cf .
@xcite , just to mention few
. basically , in a modified wos algorithm that simulates the laplace transform of the exit time and the exit location of the brownian particle , one needs to keep track of a multiplicative weight for the simulated brownian particle .
we call this algorithm the weighted walk on spheres ( wwos ) and recall it in section [ sect : wos - algorithms ] .
if the constant potential is negative , then the weight of the brownian particle can be reinterpreted as independent exponential killing of the particle .
we call this algorithm the killing walk on spheres ( kwos ) and recall it in section [ sect : wos - algorithms ] . therefore , we admit that there are already efficient algorithms for the problem studied here
. however , it is our opinion that our duffin correspondence wos algorithm ( dwos ) is of comparable efficiency to the known modified wos algorithms and has the advantage of being both simpler to comprehend and easier to implement than the modified wos algorithms known so far . indeed , dwos algorithm is simply the classical wos algorithm with an added dimension and multiplicatively modified boundary data : there is no need to keep track of any weight or killing .
moreover , if the wos algorithm is already implemented , the dwos algorithm does not need implementation : it is simply the wos algorithm with different input .
the rest of the paper is organized as follows . in section [
sect : prelim ] , we lay the setting and recall the connection between the dirichlet boundary value problems and the brownian motion . in section [ sect : duffin ] , we prove our main result , the duffin correspondence , and the stochastic representation of the solutions of the constant - potential schrdinger equation without the stopping time distribution .
section [ sect : wos - algorithms ] is devoted to the different wos algorithms and their implementations . in section [
sect : comparisons ] , we provide examples and comparisons of the different wos algorithms . finally , in section [ sect : conclusions ] we draw some some conclusions on the performance of the dwos algorithm .
let @xmath0 be a domain ( i.e. , open and connected ) satisfying assumption [ ass ] below .
let @xmath1 .
denote by @xmath2 the laplacian with respect to the variable @xmath3 .
we consider the dirichlet - type boundary value problem of the schrdinger equation with constant potential @xmath4 : @xmath5 here @xmath6 is ( continuous and ) bounded on @xmath7 .
the case @xmath8 corresponds to the yukawa equation , or the linearized poisson
boltzmann equation .
the case @xmath9 is the laplace equation .
the case @xmath10 is the helmholtz equation . the required regularity conditions for the domain @xmath11 of a helmholtz
yukawa type dirichlet boundary value problem to admit a unique bounded ( strong ) solution are best expressed by using probabilistic tools and the brownian motion : let @xmath12 be a standard @xmath13-dimensional brownian motion .
let @xmath14 be the first exit time of the brownian motion @xmath12 from the domain @xmath11 , i.e. , @xmath15 here , as always , use the normal convention that @xmath16 the following assumptions on the domain @xmath11 are always in force , although not explicitly stated later : [ ass ] 1 .
the domain @xmath11 is _ wiener regular _ ,
i.e. , @xmath17 \quad\mbox{for all } y\in\partial d.\ ] ] 2 .
the domain @xmath11 is _ wiener small _ , i.e. , @xmath18 = 1 \quad\mbox{for all } x\in d.\ ] ] 3 . finally , we assume , by using the terminology of chung and zhao @xcite , that the domain @xmath11 is _ gaugeable _ , i.e. , @xmath19 < \infty.\ ] ] [ rem : ass ] 1 . all domains with piecewise @xmath20 boundary are wiener regular .
2 . if any projection of the domain @xmath11 on any subspace @xmath21 , @xmath22 is bounded , then @xmath11 is wiener bounded .
3 . for @xmath23 ,
the gauge condition [ ass](iii ) is vacuous ; for @xmath10 it is essential .
+ actually , it follows from ( * ? ? ?
* theorem 4.19 ) that the gauge condition [ ass](iii ) is satisfied if and only if @xmath24 , where @xmath25 is the principal dirichlet eigenvalue of the negative half - laplacian , i.e. , for @xmath24 the boundary value problem @xmath26 admits only the trivial solution @xmath27 . by the rayleigh
krahn inequality @xcite @xmath28 where @xmath29 is the smallest positive zero of the bessel function @xmath30 equality is attained in if and only if @xmath11 is a ball ; e.g. for the unit ball we have @xmath31 this relation is pronounced later in formula .
numerical approximations of @xmath32 are give in table [ tab : rayleigh ] .
+ .numerical approximations of the rayleigh faber
krahn constant @xmath32 for @xmath33 . [ cols="^,^",options="header " , ] the following stochastic representation of the bounded solutions to the boundary value problem is well - known .
see , e.g. , ( * ? ? ?
* chapter 4 ) or ( * ? ? ? * chapter 4 ) : [ pro : kakutani ] the boundary value problem admits a unique bounded solution given by @xmath34.\ ] ]
associated with the helmholtz or the yukawa boundary value problem with @xmath35 , there is a classical laplace boundary value problem with @xmath9 on an extended domain @xmath36 : indeed , define @xmath37 by @xmath38 set @xmath39 and @xmath40 finally , denote @xmath41 , and set @xmath42 with this notation , consider the following family of laplace boundary value problems indexed by @xmath35 : @xmath43 [ thm : duffin ] let @xmath35 be fixed .
then @xmath44 is the unique bounded solution to the helmholtz or yukawa boundary value problem if and only if @xmath45 is the unique bounded solution to the laplace equation .
the yukawa case , @xmath8 , was shown in @xcite ( for general @xmath13 ) and in the original paper by duffin @xcite ( for @xmath46 ) .
let us consider the helmholtz case , @xmath10 .
the proof that @xmath47 if and only if @xmath48 is straightforward and can be done exactly as in the yukawa case . also , it is straightforward to see that @xmath11 satisfies assumptions [ ass](i ) and [ ass](ii ) if and only if @xmath36 satisfies assumptions [ ass](i ) and [ ass](ii ) .
the essential difference to the yukawa case is that now @xmath36 is unbounded in the @xmath49 co - ordinate .
consequently , the boundary data @xmath50 is not bounded .
this is where we need the gauge condition [ ass](iii ) .
indeed , we can approximate the solution in @xmath51 $ ] , and the result follows from the stochastic representation [ pro : kakutani ] and the dominated convergence theorem by letting @xmath52 .
let @xmath53 be an @xmath54-dimensional brownian motion and let @xmath55 then , by using the duffin correspondence theorem [ thm : duffin ] , we obtain a stochastic representation for the solution of the boundary value problem in terms of the exit location distribution only , i.e. , a stochastic solution that is independent of the exit time distribution .
[ cor : kakutani ] the boundary value problem admits a solution @xmath56.\ ] ]
if one wants to simulate the solutions of directly by using the stochastic representation of proposition [ pro : kakutani ] , then one must simulate both the exit time and the exit position of the brownian particle moving in the domain @xmath11 .
this can be done e.g. by using a weighted walk on spheres ( wwos ) , or in the yukawa setting killing walk on spheres ( kwos ) .
( it is also possible to use the recent walk on moving spheres ( woms ) algorithm developed in @xcite . ) however , by using the duffin correspondence of theorem [ thm : duffin ] one only needs to simulate the exit position of the brownian particle in the extended domain @xmath36 .
this allows one to use the classical muller s @xcite walk on spheres ( wos ) algorithm .
we call this extension the duffin walk on spheres ( dwos ) algorithm . in this section
, we will explain in detail these walk on spheres algorithms dwos , wwos , kwos for the yukawa
laplace helmholtz equations .
next , in section [ sect : comparisons ] we will provide examples and comparisons of these algorithms . we start with the most elementary problem : how to generate uniform random variables on spheres . in all the variants of the walk on spheres algorithms presented here , one needs to simulate random variables that are uniformly distributed on the @xmath57-dimensional unit spheres @xmath58 .
most mathematical software has this functionality built in . in case your favorite software does not have this functionality , algorithm [ alg : uniform - sphere ] below
explains how to generate uniform distribution on the @xmath57-dimensional unit spheres by using independent normally distributed random variables .
generating normally distributed random variables should be built - in in almost all mathematical software .
[ alg : uniform - sphere ] generate @xmath59 independent standard normal random variables .
set @xmath60 .
then @xmath61 is uniformly distributed on the @xmath57-dimensional unit sphere . for the dwos algorithm [ alg : dwos ] below
, we use the stochastic representation of corollary [ cor : kakutani ] .
recall that @xmath62 is the @xmath54-dimensional brownian motion , @xmath63 and @xmath64 are the probability law and expectation , respectively , under which @xmath65 . under the probability law @xmath63 , denote @xmath66 [ rem : exit - first ] @xmath67 and @xmath68 are independent , and @xmath69 for @xmath70 , denote @xmath71 .
the stochastic approximation for @xmath72 is @xmath73 here@xmath74 is the termination - step of the trajectory of the each individual brownian particle @xmath75 starting from @xmath76 : @xmath77 for each @xmath78 , and @xmath79 , @xmath70 , the individual particle exit locations @xmath80 for the approximating sum are generated by duffin walk on spheres ( dwos ) algorithm [ alg : dwos ] below : [ alg : dwos ] fix a small parameter @xmath81 .
1 . initialize : @xmath82 .
2 . while @xmath83 : 1 . set @xmath84 .
sample @xmath85 independently from the unit sphere @xmath86 ( by using algorithm [ alg : uniform - sphere ] ) .
3 . set @xmath87 .
3 . when @xmath88 : 4 .
set @xmath89 to be the orthogonal projection of @xmath90 to @xmath91 .
return @xmath89 .
suppose we want use the wos algorithm directly without the duffin correspondence . to do this , we must estimate the term @xmath92 in proposition [ pro : kakutani ] .
suppose the wos algorithm takes @xmath93 steps to hit the boundary with balls of radii @xmath94 .
the weighted walk on spheres ( wwos ) algorithm [ alg : wwos ] below is based on the fact that the term @xmath92 can be decomposed into independent terms @xmath95 where the @xmath96 s are the exit times of the brownian motion from balls of radius @xmath97 , and these exit times are also independent of the exit locations from the ball . consequently
, at each step @xmath98 of the wos algorithm , the brownian particle gains ( or loses ) an independent multiplicative weight that is given by @xmath99.\ ] ] here @xmath100 . by using the @xmath101-self - similarity of the brownian motion
we see that @xmath102 = { \mathbb{e}}\left[e^{-\lambda r^2 \tau_{1}}\right ] = \psi(\lambda r^2).\ ] ] for @xmath103 the function @xmath104 is well - known , cf .
wendel @xcite : @xmath105 here @xmath106 is the modified bessel function @xmath107 for @xmath108 , as far as we know , no simple formula for @xmath109 is known . however , the distribution function of @xmath110 is well - known , cf .
kent @xcite or ciesielski and taylor @xcite : @xmath111 = 1 - \frac{1}{2^{\nu-1}\gamma(\nu+1)}\sum_{i=1}^{\infty } \frac{j_{\nu , i}^{\nu-1}}{j_{\nu+1}(j_{\nu , i})}\exp\left\{-\frac12 j_{\nu , i}^2 \,t\right\}.\ ] ] here @xmath112 , @xmath113 , and @xmath114 s are the positive zeros of the bessel function @xmath115 given by in the increasing order . consequently , by simple substitution to combined with change of differentiation and summation , @xmath116 \nonumber \\ & = & \frac{-1}{2^{\nu-1}\gamma(\nu+1)}\sum_{i=1}^{\infty } \frac{j_{\nu , i}^{\nu-1}}{j_{\nu+1}(j_{\nu , i})}\int_0^\infty e^{-\mu t}\,{\mathrm{d}}\left [ \exp\left\{-\frac{1}{2 } j_{\nu , i}^2\ , t\right\ } \right ] \nonumber \\ & = & \frac{1}{2^{\nu}\gamma(\nu+1)}\sum_{i=1}^{\infty } \frac{j_{\nu , i}^{\nu+1}}{j_{\nu+1}(j_{\nu , i})}\int_0^\infty \exp\left\{-\left(\mu+\frac{1}{2 } j_{\nu , i}^2\right ) t\right\ } \ , { \mathrm{d}}t .
\label{eq : psi}\end{aligned}\ ] ] by the gauge condition @xmath117 . therefore , @xmath118_{t=0}^{\infty } \\ & = & \frac{1}{2^{\nu}\gamma(\nu+1)}\sum_{i=1}^{\infty } \frac{j_{\nu , i}^{\nu+1}}{j_{\nu+1}(j_{\nu , i})\left(\mu+\frac12 j_{\nu , i}^2\right)}.\end{aligned}\ ] ] consequently , we have obtained @xmath119 a formula that is true for all @xmath120 , i.e. , whenever the gauge condition [ ass](iii ) for the unit ball holds . while it is possible to directly use the formula in numerical computations ,
our experimental results show that this is very inefficient as the sum in the formula converges very slowly and several thousand terms are needed for workable accuracy .
there are at least two reasonable approaches to address this shortcoming : 1 .
one may use the formula to tabulate values of the function @xmath121 $ ] for given values of @xmath122 , where @xmath123 and @xmath124 , and then use the identity and the elementary properties of the expected value to obtain the approximation @xmath125-{\mathbb{p}}\left[\tau_1\le t_j\right]\right).\ ] ] 2 .
one may use the formula to tabulate values of the function @xmath126 for various values of @xmath127 , and then use the table and appropriate interpolation to approximate the function @xmath126 for an arbitrary value of @xmath128 .
the first approach was used in examples of section [ sect : comparisons ] .
approximations for function @xmath126 , for various values of @xmath129 and @xmath13 are illustrated in figure [ fig : psifun ] ( cf .
the figure in @xcite for an illustration of @xmath104 for positive values of @xmath130 ) . for @xmath131 for @xmath132 , where @xmath133 on the top ( left ) .
the function @xmath134 , @xmath127 , where the dimension are @xmath132 , where @xmath133 on the top ( right ) .
, title="fig:",width=207 ] for @xmath131 for @xmath132 , where @xmath133 on the top ( left ) .
the function @xmath134 , @xmath127 , where the dimension are @xmath132 , where @xmath133 on the top ( right ) . ,
title="fig:",width=207 ] note that is the extension of the @xmath104-function , that appears in @xcite , but with different parametrization .
now we are ready to give the weighted walk on spheres ( wwos ) algorithm : the approximation for @xmath72 is , by proposition [ pro : kakutani ] , @xmath135 here@xmath136 is the exit time for the each individual particle and @xmath137 .
the individual particle exit locations @xmath138 and weights @xmath139 are generated by algorithm [ alg : dwos ] below : [ alg : wwos ] fix a small parameter @xmath81 .
1 . initialize : @xmath140 , @xmath141 2 . while @xmath142 : 1 .
set @xmath143 .
sample @xmath85 independently from the unit sphere @xmath144 ( by using algorithm [ alg : uniform - sphere ] ) .
3 . set @xmath145 and @xmath146 .
here @xmath104 is given by for @xmath8 and for @xmath10 .
3 . when @xmath147 : 4 .
set @xmath148 to be the orthogonal projection of @xmath149 to @xmath7 .
return @xmath148 and @xmath150 . for the yukawa case @xmath8
, the weight loss @xmath151 of the particle can be interpreted as independent exponential killing of the particle .
see @xcite , @xcite or @xcite for details .
consequently , the wwos algorithm [ alg : wwos ] can be reinterpreted as killing walk on spheres ( kwos ) .
our estimator for @xmath72 is @xmath152 where @xmath153 , @xmath78 are independent simulations of the trajectories brownian particles starting from point @xmath154 , and the set @xmath155 contains the particles that are not killed ; @xmath136 is the termination - step time of the algorithm .
the individual particles are generated by algorithm [ alg : kwos ] below .
[ alg : kwos ] fix a small parameter @xmath81 .
1 . initialize : @xmath140 .
2 . while @xmath142 : 1 . set @xmath143 .
2 . kill the particle with probability @xmath156 .
if the particle is killed , the algorithm terminates and returns @xmath157 .
sample @xmath85 independently from the unit sphere @xmath144 ( by using algorithm [ alg : uniform - sphere ] ) .
4 . set @xmath145 .
3 . when @xmath147 : 4 .
set @xmath148 to be the orthogonal projection of @xmath149 to @xmath7 .
return @xmath148 .
in this section , we give examples to motivate our algorithms and to illustrate their potential applications .
the examples were computed by using a straightforward implementation algorithms in one and two - dimensional settings and chosen from the point of view of visualization .
all computations were performed on a macbook air laptop .
wolfram mathematica 10.2 and basic implementations of the algorithms with no performance optimizations were used .
obviously , the stochastic approaches presented here are more attractive in higher dimensions , where many deterministic simulation methods are not available , or lead into excessive computation times .
also , it should be noted that the algorithms in this paper are particularly suitable for parallel computation , as simulated paths are independent from each other .
let @xmath158 $ ] and consider the differential equation @xmath159 with boundary values @xmath160 and @xmath161 .
then the exact solution to the boundary value problem is given by @xmath162 the exact solution @xmath44 as well as its approximations with the dwos algorithm , where the problem is first lifted into dimension two by using the duffin correspondence , and with the wwos algorithm are illustrated in figure [ fig : onedim ] .
( dashed ) and its approximations @xmath163 with dwos ( left ) and wwos ( right ) algorithms .
, title="fig:",width=207 ] ( dashed ) and its approximations @xmath163 with dwos ( left ) and wwos ( right ) algorithms . , title="fig:",width=207 ] next , we consider two simple boundary value problems of the equation on polygonal domains in the plane : 1 .
the trapezoidal domain @xmath164 defined by the points @xmath165 and @xmath166 , where the boundary values are given by @xmath167 and with @xmath168 ( cf .
example 4.4 and figure 2 of @xcite ) .
2 . the non - convex l - shaped domain @xmath169 defined by the points @xmath170 and @xmath166 , where boundary values are given by the non - continuous function @xmath171 and @xmath172 .
we compute an approximation @xmath173 to the solution of the above boundary value problems by using both the dwos algorithm and the wwos algorithm . for dwos ,
the problem is lifted to the dimension three .
the dwos and wwos on the domain @xmath169 are illustrated in figure [ fig : wos23diml ] .
approximations to the solutions are illustrated in figure [ fig : sol - trape ] and figure [ fig : sol - l ] .
our experiments suggest that wwos algorithm is more stable than dwos and thus a smaller number of simulations are required for a comparable result . with dwos ( left ) and wwos ( right ) algorithms . in dwos , the problem is first lifted into the dimension three by using the duffin correspondence .
the wwos algorithm is purely two - dimensional , but it requires computation of the weight function @xmath104 . , title="fig:",width=207 ] with dwos ( left ) and wwos ( right ) algorithms . in dwos ,
the problem is first lifted into the dimension three by using the duffin correspondence .
the wwos algorithm is purely two - dimensional , but it requires computation of the weight function @xmath104
. , title="fig:",width=207 ] ( dwos , left ) and @xmath174 ( wwos , right ) and of the solution on the domain @xmath164 .
the boundary values are given by the function @xmath175 and @xmath168 . , title="fig:",width=219 ] ( dwos , left ) and @xmath174 ( wwos , right ) and of the solution on the domain @xmath164 .
the boundary values are given by the function @xmath175 and @xmath168 . , title="fig:",width=196 ] ( dwos , left ) and @xmath174 ( wwos , right ) and of the solution on the domain @xmath169 .
the boundary values are given by the function @xmath176 and @xmath177 . , title="fig:",width=219 ] ( dwos , left ) and @xmath174 ( wwos , right ) and of the solution on the domain @xmath169 .
the boundary values are given by the function @xmath176 and @xmath177 .
, title="fig:",width=196 ]
it is clear that the kwos algorithm is better than the wwos algorithm , when it is applicable ( i.e. , in the yukawa case ) .
our experiments in section [ sect : comparisons ] suggest that , at least on low dimensions , wwos algorithm is more stable than the dwos algorithm . consequently , it seems that the dwos algorithm is best used in high dimensions , where adding one extra dimension should not make much difference .
however , the dwos algorithm is an extension , not a modification , of the classical wos algorithm , i.e. , if one has wos already implemented , implementing dwos is simply a matter of giving different input parameters to the wos algorithm . | we show that a constant - potential time - independent schrdinger equation with dirichlet boundary data can be reformulated as a laplace equation with dirichlet boundary data . with this reformulation , which we call the duffin correspondence ,
we provide a classical walk on spheres ( wos ) algorithm for monte carlo simulation of the solutions of the boundary value problem .
we compare the obtained duffin wos algorithm with existing modified wos algorithms . |
evidences accumulated in the past decade point to the phenomenon that feedback from active galactic nuclei ( agns ) plays a crucial role in galaxy formation and evolution .
outflows in agns connect the central supermassive black holes ( smbhs ) to their host galaxies , and regulate their co - evolution ( granato et al .
2004 ; scannapieco & oh 2004 ; hopkins et al . 2008 ; also see antonuccio - delogu & silk 2010 for a recent review ) .
observationally , the most direct and obvious performance of agn outflows is the high - speed blueshifted broad absorption lines ( bals ) . because of the high fraction in optically selected quasars ( 10%-20% ) , high - ionization bal ( hibal ) quasars
are widely studied , with strong absorption troughs in high ionization ions such as , , , and , up to a velocity of @xmath11 ( e.g. , weymann et al .
1991 ; trump et al . 2006 ; gibson et al .
bals are also detected occasionally ( @xmath12 in bal quasars and @xmath13 in all quasars ) in low ionization species such as and , named as low - ionization bals ( lobals ) ( e.g. , weymann et al .
1991 ; reichard et al . 2003 ; zhang et al .
moreover , felobal quasars have been observed to exhibit the broad absorption troughs in and/or , which are rare and only found in only @xmath12 of lobal quasars ( hall et al .
2002 ) .
statistical studies are widely used to explain the bal phenomenon .
bal quasars on average have red continua ( weymann et al .
1991 ; richards et al . 2003 ; zhang et al . 2010 , 2014 ) weak x - ray emission ( e.g. , green et al .
1995 ; brinkmann et al .
1999 ; wang et al . 1999
; brandt et al . 2000 ; gallagher et al . 2002 , 2006 ; fan et al . 2009 ) , are more frequently detected in quasars with higher eddington ratio and higher luminosity ( ganguly et al .
2007 ; zhang et al . 2010 , 2014 ) .
furthermore , the outflow velocity and strength are tied to the properties of quasars . observed outflow velocities increase with the blueness of uv spectral slope , the enhancement of black hole accretion and the equivalent width of emission ( hamann 1998 ; laor & brandt 2002 ; gungly et al .
2007 ; misawa et al . 2007 ; baskin et al . 2013 ; zhang et al . 2014 ) .
in particular , the minimum velocity of absorption is even more strongly correlated with uv spectral slope than the maximum velocity ( zhang et al .
these findings indicate the primary role of radiatively driven winds in the outflow phenomenon , and the importance of the spectral energy distribution shape ( sed ) in governing the dynamics of outflows .
meanwhile , all outflow parameters dramatically and monotonically increase with hot dust emission .
these correlations can be more naturally interpreted as the dusty outflow scenario rather than the dust - free outflow scenario , where the dust is intrinsic to the outflows or interaction with torus clouds ( wang et al .
2013 ; zhang et al .
2014 ) . on the other hand , the study of variation of bal troughs and the associated rare absorption systems , i.e. , lobals , and unusual bals ( mostly felobals ) can provide a new and more effective perspective to understand the physical conditions , locations and origins of the absorbers , and constraint on the outflow mechanism , which gradually becomes a hot topic of the bal research ( e.g. , hall et al .
2002 , 2011 ; zhou et al . 2006 ; lundgren et al . 2007 ; gibson et al . 2008 , 2010 ; krongold et al .
2010 ; zhang et al .
2011 , 2015a , 2015b ; capellupo et al . 2012 ; vivek et al .
2012 , 2014 ; filiz ak et al . 2013
; welling et al .
currently , one of the rarest known bals is non - stellar balmer line absorption .
it has previously been reported only in nine objects as follows : ngc 4151 ( hutchings et al . 2002 ) , sdss j112526.12 + 002901.3 ( hall et al .
2002 ; shi et al . in pre .
) , sdss j083942.11 + 380526.3 ( aoki et al .
2006 ) , sdss j125942.80 + 121312.6 ( hall 2007 ; shi et al . in pre .
) , sdss j102839.11 + 450009.4 ( wang et al . 2008 ) , sdss j172341.10 + 555340.5 ( aoki 2010 ) , lbqs 1206 + 1052 ( ji et al . 2012 ) , sdss j222024.59 + 010931.2 ( ji et al . 2013 ) and sdss j112611.63 + 425246.4 ( wang & xu 2015 ) .
however , strictly speaking , more than half of them can not be classified as balmer bals based on the criteria for bals ( see table 1 of zhang et al .
( 2010 ) for a summary and comparison ) , as their absorption widths are narrower than 1000 km s@xmath4 . in this paper
, we report a quasar ( sdss j152350.42 + 391405.2 , hereafter 1523 ) with a emission redshift of @xmath14 .
this object shows the broadest and highest blueshifted velocity balmer bals known so far in the quasars , suggesting the strong , high speed and high column density outflow materials in the nuclear region .
the organization of this paper is as follows .
the data we used will be described in section 2 .
we will fit the spectrum and analyze the balmer bals in section 3 , and discuss the properties and possible origins of bals in section 4 .
a summary of our results will be given in section 5 . throughout this paper ,
we adopt the cold dark matter ` concordance ' cosmology with h0 = 70 km s@xmath15mpc@xmath15 , @xmath16 , and @xmath17 .
cccc & @xmath18 mjy & 1994/08/13 & first + 20 cm & @xmath19 mjy & 1998/04/16 & nvss + @xmath20 & @xmath21 mag & 2003/02/11 & sdss + @xmath22 & @xmath23 mag & 2003/02/11 & sdss + @xmath24 & @xmath25 mag & 2003/02/11 & sdss + @xmath26 & @xmath27 mag & 2003/02/11 & sdss + @xmath28 & @xmath29 mag & 2003/02/11 & sdss + @xmath30 & @xmath31 mag & 1999/05/24 & 2mass + @xmath7 & @xmath32 mag & 1999/05/24 & 2mass + @xmath33&@xmath34 mag & 1999/05/24 & 2mass + @xmath35 & @xmath36 mag & 2010/01/19 , 07/17 , 07/22 & wise + @xmath37 & @xmath38 mag & 2010/01/19 , 07/17 , 07/22 & wise + @xmath39 & @xmath40 mag & 2010/01/19 , 07/16 , 07/22 & wise + @xmath41 & @xmath42 mag & 2010/01/19 , 07/16 , 07/22 & wise + @xmath43 & & 2005/04/202013/09/28 &
catalina + @xmath43 & & 2015/04/20,05/13 & bsst 1523 is very bright with a galactic extinction corrected magnitude of 16.59 at @xmath24-band .
the first ( the faint images of the radio sky at twenty cm , becker et al .
1995 ) and nvss ( nrao vla sky survey ; condon et al .
1998 ) surveys show that there is no radio variation at 1.4 ghz , with the peak flux of @xmath44 and @xmath45 mjy respectively . at the high energy band ,
its x - ray emission is very weak and not detected by xmm - newton and chandra x - ray observatories .
the sed from ultraviolet ( uv ) to middle - infrared ( mir ) , comes from the data taken with the sloan digital sky survey ( sdss , york et al .
2000 ) , the two micron all sky survey ( 2mass ; skrutskie et al . 2006 ) and the wide - field infrared survey explorer ( wise ; wright et al .
the uv emission is attenuated by the strong and absorption . meanwhile ,
this object is also monitored by the catalina surveys ( drake et al .
2014 ) for the nine years ( 278 epochs ) , beginning in april 2005 .
figure [ f1 ] presents the @xmath43-band light curve of 1523 ( green squares ) , which shows very weak long - term variability with large measurement errors . the antarctic bright star survey telescope ( bsst ; tian et al .
, in pre . )
also gives 8 epoch photometric observations in the @xmath43-band at the lijiang observational station of yunnan observatories .
the latest monitoring result ( red square ) confirms there is no apparent change in the optical continuum for 1523 .
the photometric data are summarized in table [ tab1 ] .
the optical spectrum of 1523 was first taken with the 3.5 m telescope on apache point observatory ( apo ) at may 26 , 1997 in the first bright quasar survey ( fbqs , white et al .
the apo spectrum has a wavelength coverage from 3650 to 10000 at @xmath46 resolution .
1523 was also considered as a quasar candidate from the spectroscopy taken with the sdss 2.5 m telescope on may 6 , 2003 and may 28 , 2012 .
the first observation was recorded into the sdss data release 5 ( dr5 ; adelman - mccarthy et al .
two 5400s exposures were taken with the original sdss spectrographs , which provide a high signal - to - noise ( snr ) spectrum at the resolution @xmath47 and the wavelength coverage from 3800 to 9200 ( stoughton et al .
the second observation was taken in the baryon oscillation spectroscopic survey ( boss ; dawson et al .
2013 ) and pubilshed in the sdss tenth data release ( dr10 , ahn et al . 2013 ) , the boss spectrum has a wider wavelength range covering 361 nm - 1014 nm with a resolution of 1300 at the blue side and 2600 at the red side , respectively .
this instrument is fed by smaller optical fibers , each subtending 2 on the sky ( smee et al .
2013 ) .
the near - infrared ( nir ) spectrum of 1523 was taken with the triplespec spectrograph of hale 200-inch telescope ( p200 ) at palomar observatory on april 16 , 2012 .
1523 was exposed four times , each for 240 seconds .
triplespec ( wilson et al . 2004 ) is a cross - dispersed nir spectrograph that provides simultaneous wavelength coverage from 0.9 to 2.46 microns at a resolution of 1.4 - 2.9 .
the raw data were processed using idl - based spextool software(vacca et al .
2003 and cushing et al .
there are two gaps in infrared spectrum around 1.35 microns and 1.85 microns due to the atmosphere transmissivity .
fortunately , the redshifted h@xmath0 emission line is detected with the triplespc at @xmath30-band .
since the time interval between the sdss dr10 and p200 triplespec observations is just 12 days in observed frame , the data are considered as quasi - simultaneous which we will use in the following analysis .
after corrected for the galactic reddening of @xmath48 = 0.021 mag ( schlegel et al . 1998 ) , we transformed the photometric data as well as optical and nir spectrum into the rest frame with its emission redshift , which are shown in red squares , black and blue curves in panle ( a ) of figure [ f2 ] , respectively .
the triplespec spectrum , after correcting for the aperture effect , agrees very well with the 2mass photometric data .
the sdss and triplespec continuum spectra and wise photometric data suggest two basic components .
one is a singel power - law , @xmath49 , representing the nuclear spectrum from the accretion disk ( the pink dashed line ) .
it is estimated from continuum windows ( [ 3790 , 3810] , [ 5600 , 5630] , [ 6950 , 7000] ) that are not seriously contaminated by emission lines ( e.g. , forster et al .
another is the nir emission bump at @xmath37-band corresponding to the blackbody radiation from the hot dust with @xmath50 k ( green dashed line ) .
1523 is a typical bal quasar , and first classified as a radio - detected bal quasar with @xmath51 km s@xmath4 by becker et al .
it is also contained in the sdss dr5 low - redshift bal quasar sample ( zhang et al . 2010 ) .
the measured maximum and minimum velocities of bal trough are -20,000 and -11,500 km s@xmath4 , and the absorption index @xmath52 is 3227 km s@xmath4 .
we note that the absorption intensity of bal listed in the above literature is likely underestimated as the absorbing troughs of and uv62 overlap each other , and can not be discriminated easily . in zhang
et al . ( 2015b ) , 1523 was reidentified to be an ` overlapping - trough ' felobal quasar because of almost no continuum windows below mg ii and causing overlapping absorption troughs . in panel ( b ) of figure [ f2 ]
, we show the spectral details of 1523 from 2300 to 7000 .
several regions of the spectrum are absorbed and significantly below the nuclear continuum . except bals of and
, balmer bals can be also found by visual examination .
we further selected two quasars ( sdss j082806.18 + 063608.3 and sdss j125807.45 + 232921.6 ) using the pair - matching method ( zhang et al .
2014 ; liu et al .
2015 ) , which can match the continuum slop and emission multiplets of 1523 except for the absorbing troughs . compared with the spectra of the two quasars ( green and red curves ) , bal troughs of h@xmath0 and h@xmath53 are evident .
we marked the potential bal absorption regions by gray dashed lines . in order to exactly measure the absorption troughs of balmer lines
, we simultaneously fit the continuum and emission lines in the h@xmath53 and h@xmath0 regions using the code of dong et al .
( 2008 ) . in brief , the optical continuum from 4000 to 7000 is approximated by a single power - law ( @xmath54 ) .
the value of the slope estimated from continuum windows is took as initial values of @xmath53 .
indeed , we also try to use a broken power - law with a break wavelength of 5600 , i.e. , @xmath55 for the h@xmath53 region and @xmath56 for the h@xmath0 region .
however , we get the almost same values of @xmath57 and @xmath58 .
that means a single power - law is enough to model the continuum .
multiplets , both broad and narrow , are modeled using the template provided by vron - cetty et al .
emission lines are modeled as multiple gaussians : two gaussians for broad balmer lines , one gaussian for [ ] .
there are no significant narrow emission lines in the spectra of 1523 , thus we do not add gaussians for narrow emission lines .
additionally , we assume that the [ ] @xmath594959,5007 doublets have the same redshift and profile and their flux ratio is fixed to theoretical value .
we notice that the spectrum of sdss j125807.45 + 232921.6 can cover the blue wing ( @xmath60 km s@xmath4 ) and red wing of broad h@xmath0 line in 1523 ( figure [ f2 ] , panel ( b ) ) , that means h@xmath0 emission is likely unaffected by the bal absorption .
thus , we just mask the potential bal absorption regions ( gray dashed lines ) in the fitting . in top panels of figure [ f3 ]
, we show the rest - frame spectra of the sdss and triplespec in black curves and our best - fit models in red curves .
a more accurate continuum around h@xmath0 and h@xmath53 , @xmath61 , is shown in pink dashed line .
the blue solid lines show the broad and narrow components of optical emission and the green curves represent the three strong balmer lines . in panel
( a ) , the strong narrow 37 multiplets are present in h@xmath53 bal trough .
this implies that the outflow winds may only obscure the nuclear continuum and broad emission lines , but not the narrow emission lines .
this speculation is consistent with our detailed calculation in 4.2 .
furthermore , it can be seen from the normalized spectra of h@xmath0 , h@xmath53 and h@xmath1 ( figure [ f3 ] , panel ( c ) ) that the h@xmath0 trough is polluted by the sky lines .
the latter is plotted as the grey line in panel ( b ) of figure [ f3 ] for comparison .
that is more remarkable in the normalized spectrum of h@xmath0 in velocity space , the corresponding polluted velocity regions are marked by green lines ( panel ( c ) ) . the ` true ' normalized spectrum in these polluted regions is approximately given through interpolation of those in unpolluted regions .
absorption parameters of balmer bal troughs are listed in table [ tab2 ] .
lccc wavelength( ) & 6564.41 & 4862.68 & 4341.68 + @xmath62 @xmath63 & 0.6400 & 0.1190 & 0.0446 + @xmath64@xmath65 ( km s@xmath4 ) & 16,821 & 16,109 & 13,722 + @xmath66@xmath67 ( km s@xmath4 ) & 5285 & 4095 & 8131 + @xmath52 @xmath68 ( km s@xmath4 ) & [email protected] & [email protected] & - + observed depth@xmath70 ( % ) & 33 & 19 & 5 + model depth@xmath70 ( % ) & 33.9 & 18.1 & 7.6 as one know , emission is unique in individual quasars and not exactly the same as the template .
furthermore , the ratio of multiplets with different excitation is very sensitive to the temperature and density of the gas .
the balmer bal troughs are unfortunately falling into the regimes of multiplets .
thus , we used a new template derived from the keck spectrum of iras 07598 + 650 ( vron - cetty et al . 2006 ) to investigate the impact of templates .
we first decompose the continuum emission of iras 07598 + 650 from the keck spectrum trough a single power - law fitting based on the above - mentioned continuum windows in 2 .
after subtracting the modeled continuum , the residual spectrum which includes broad and narrow and balmer emission lines is used as emission template . in panel
( a ) and ( b ) of figure [ f3 ] , the cyan curves show the new best - fit model , which is the sum of power - law continuum ( pink ) and scaled emission template .
it can be seen that the new unabsorbed model around h@xmath0 and h@xmath53 bal troughs is below the best - fitting template causeing weaker h@xmath0 and h@xmath53 absorption .
conversely , it shows a stronger h@xmath1 bal than the previous measurement . in order to further rule out the possibility that balmer bals are resulting from the artifacts of spectral fittings
, we constructed a sample of 500 quasars whose spectra can match the slop and primary multiplets , i.e. , @xmath59 4472 - 4731 and @xmath59 5169 - 5325 of 1523 .
we used them to normalize the observed spectra of 1523 and calculated the absorption equivalent width of balmer bals .
the average values are @xmath71 km s@xmath4 , @xmath72 km s@xmath4and @xmath73 km s@xmath4for h@xmath0 , h@xmath53 and h@xmath1 bals respectively . from the average values and dispersion of the absorption equivalent widths
, it can be seen that the uncertainty of templates and/or the continuum determination can affect the measurement of bal parameters , but the existence of balmer bals is unquestionable .
percentage absorption depths of the troughs ( table [ tab2 ] ) decrease as the upper term of the transition increases .
however , the decline in depth is less than the decrease in transition oscillator strengths ( @xmath74 from h@xmath0 to h@xmath53 ) , leading to the absorption saturated .
there is a residual intensity of @xmath75 of the modelled flux at the velocity of the maximum absorption depth ( figure [ f3 ] , panel ( c ) ) , suggesting that the absorption materials only obscure @xmath76 of the total continuum region .
theoretically , the absorption depth is defined as @xmath77 for a partially obscured absorber , where @xmath78 is the normalized intensity in the troughs , @xmath79 is percentage covering factor of the absorber and @xmath80 is the optical depth for the relevant ion ( e.g. , hall et al .
we can calculate @xmath81 for each balmer trough from the normalized spectrum , and the relative values of @xmath82 are determined by the known oscillator strengths .
, where @xmath83 is the ratio of the optical depth of h@xmath53 to that of h@xmath0 , the @xmath84 are the statistical weights , the @xmath85 are the oscillator strengths , and the @xmath86 are the wavelengths of the lines .
] thus , we can estimate @xmath79 and @xmath80 through the observed h@xmath0 and h@xmath53 troughs .
the maximum absorption depth of the balmer lines is found to be @xmath87 with coveing factor @xmath88 at the velocity of maximum depth ( table [ tab2 ] ) . in panel
( d ) of figure [ f3 ] , the covering factor @xmath79 and the optical depth of h@xmath0 absorption @xmath80 are shown as a function of the blueshiftted velocity .
the absorption materials have the maximum covering factor at the velocity of the maximum absorption depth ( @xmath89 km s@xmath4 ) and smaller covering factors at the higher or lower velocities .
the optical depths , ie . , the column densities , decrease as a function of the blueshifted velocity from @xmath90 at @xmath91 km s@xmath4and @xmath92 at @xmath93 km s@xmath4 . in panel(c ) of figure [ f3 ] , we compared the theoretical absorption troughs ( blue curves ) with the observed normalized spectra for balmer bals .
we find that the observed absorption trough for h@xmath1 is shallower than the theoretical profile .
the absorption of h@xmath1 is relatively weak and difficult to accurately measure . using the optical depths derived above
, we calculate the column densities of h i as a function of velocity using the general expression ( e.g. , arav et al .
2001 ) @xmath94 where @xmath95 and @xmath96 are the wavelength and the oscillator strength of h@xmath0 , respectively .
the results are shown in panel ( d ) of figure [ f3 ] .
the total column density is obtained @xmath97 @xmath98 by integrating equation [ colden ] .
only the blue wing of the redshifted * @xmath610830 emission line is detected with the triplespec spectrum at h - band ( panel ( a ) of figure [ f2 ] ) , to obtain the unabsorbed spectral intensity around * @xmath610830 , we used the following procedure .
we downloaded the high - quality near - infrared broad emission line spectra presented by landt et al .
( 2008 ) and found four objects which can be compared with 1523 : 3c 273 , iras 1750 , he 1228 + 013 and pds 456 .
these spectra are similar to that of 1523 in terms of emission lines , but different in their continuum slopes .
thus we could use line - free spectral regions in each comparison object to determine the continuum and thus the normalized emission line profiles .
we focus on a limited wavelength range between 0.7 and 1.5 @xmath99 m and fit the line - free continuum bands of each object with a single power - law for the accretion disk and a black - body emission component for hot dust emission .
then the emission line templates are obtained by dividing the observed spectra by the best - fitting continuum . finally , we multiplied the emission line templates by a single power - law and black - body emission to reconstruct the nir spectrum of 1523 .
irac 1750 provided the best - matched nir emission line template with the minimized @xmath100 . in figure
[ f4 ] , left panel shows the observed spectra of * @xmath610830 regime overplayed with the best - fit model for 1523 .
the pink and orange lines represents the power - low and black - body emission components , respectively . for comparison , we overplotted the emission line template multiplied by a power - law component for irac 1750 ( blue curve ) .
right - bottom panel shows the normalized spectrum of * bal trough obtained by dividing the observed spectrum by the best - fitting emission line plus continuum template .
trough the comparisons of * and balmer bals , the velocity structure of * bal ( black curve ) is similar to those of balmer bals .
table [ tab3 ] gives outflow velocities of the absorber for various balmer bals , as reported in literature . in three cases
( sdss j125942.80 + 121312.6 , lbqs 1206 + 1052 and sdss j222024.59 + 010931.2 ) , the absorption widths of balmer absorption troughs are @xmath101 km s@xmath4 , and can be classified as balmer bals . however , the widths for other quasars are only several hundred km s@xmath4 , and could be classified as balmer nals . as shown in table [ tab2 ] , the widths of balmer bals in 1523 are @xmath102 km s@xmath4 , and are the broadest balmer absorption lines which have ever been found .
the redshift of the balmer absorption troughs in 1523 is @[email protected] , and the blueshifted velocity is @xmath3 km s@xmath4 , which can even reach @xmath104 km s@xmath4 with regard to the balmer emission lines .
it is a factor of two higher than the maximum blueshifted velocity from the previously known balmer absorption lines .
in fact , the blueshifted velocities in two - thirds of the known balmer bals are very small ( only @xmath105 km s@xmath4 ) . compared with the above two balmer bal quasars with strong [ ] emission in literature , we find that the [ ] line in 1523 is relatively weak ( ) .
1523 also has the weakest [ o iii ] emission among nine agns with balmer absorption lines .
1523 and sdss j222024.59 + 010931.2 are obviously inconsistent with the previous assertion that balmer bals are found in felobal quasars with relatively strong [ ] emission ( aoki et al .
2006 ; hall et al .
2008 ) .
among the ten objects in literature and this work , absorption lines are usually detected in * and * .
* absorption lines arise from the metastable triplet level * @xmath106 , which is populated by recombination from he@xmath107 with electrons .
transitions from this level will generate a series of absorption lines at @xmath593189 , 3889 , 10830 , which are detected in seven quasars , except for sdss j112611.63 + 425246.4 , sdss j125942.80 + 121312.6 and sdss j172341.10 + 555340.5 . another
interesting fact is that seven out of ten balmer absorption agns show abundant absorption lines arising from the excited the co - occurrence may indicate that it is probable that these three absorption phenomena are closely related .
combined diagnostics of them can determine the density and the ionization state of the absorption gas , and put constraints on the geometry and physical conditions of outflows .
for example , the absorbers in sdss j112526.12 + 002901.3 are considered to have the parameters of @xmath108 , @xmath109 and @xmath110 .
the derived distance from the central engine is about @xmath111 pc , which is about 10 times the size of broad emission line region and similar to the radius of the inner edge of dusty torus ( shi et al .
, in pre . ) .
these estimations provide us some enlightenment about the physical properties of the absorbers in 1523 .
1523 has the strongest balmer bal troughs discovered to date , and exhibits simultaneously the broad absorption lines of * @xmath610830 and both ground and excited state .
however , these absorption lines have extremely broad velocity structures and heavily overlap each other , which can not be trivially separated . since the absorption lines are blended , it is not trivial to obtain the geometry and physical conditions of outflow winds via the combination of the absorption line diagnostics .
the large - scale synthesis code cloudy ( c10.00 ; ferland et al .
1998 ) is employed to evaluate the absorption lines of these atoms / ions in the extensive parameter space .
the geometry is assumed as a slab - shaped absorbing medium exposed to the ionizing continuum from the central engine with uniform density , metallicity and abundance pattern .
the full 371 levels fe@xmath107 model is used to reproduce the absorption , and solar elemental abundance is adopted and the gas is assumed free of dust .
in addition , a typical agn multi - component continuum is setted as incident ionizing radiation , where the big bump " component peaks at @xmath112 1 ryd and is parameterized by t=@xmath113 k. the x - ray to uv ratio is @xmath114 , and the low - energy slope of the big bump continuum is @xmath115 which is measured in 3.1 .
the slope of the x - ray component is setted to the default @xmath116 ( see details in hazy , a brief introduction to cloudy c10 ; http://www.nublado.org ) .
we calculated a series of photoionization models with different ionization parameters , electron densities and hydrogen column densities .
the ranges of parameters are @xmath117 , @xmath118 and @xmath119 with a step of 0.5 dex .
synthetic model spectra are constructed to compare the simulations with observation .
the underlying assumption is that other absorptions such as @xmath120 and have the same profile as balmer lines , which means for any ion the fraction of column density at given radial velocity to the integrated ionic column density is the same , and the covering factor as function of radial velocity is the same .
therefore in constructing the model spectra , for a given absorption line , the ionic column density predicted by cloudy on the lower level of the transition is distributed to different outflow velocities following the fractional column density distribution versus @xmath121 from balmer series , evaluating the optical depth @xmath122 as function of @xmath121 . and then considering the effect of partial covering as @xmath123 to get the model absorption profile , where @xmath124 is the template for unabsorbed background radiation field .
when comparing the model spectrum ( @xmath125 ) with the observed one , the spectrum from 4000 to 7000 covering the three balmer bals are employed in the fitting process .
@xmath124 is actually the unabsorbed intrinsic spectrum , for example , the red curves shown in top panels of figure [ f3 ] for balmer lines . for each set of parameters ( @xmath126 , @xmath127 and @xmath128 ) , we calculated all possible combinations of individual models , and selected the one with minimized @xmath100 which is plotted in panel ( a ) and ( b ) of figure [ f5 ] ( green curves ) .
we derived physical parameters @xmath9 , @xmath8 and @xmath129 for balmer bals .
the green curves represent the model profiles of h@xmath53 and h@xmath0 absorption which match well to the observed spectra .
meanwhile , we also plotted the model absorption trough of @xmath130 in panel ( c ) of figure [ f5 ] .
the model absorption trough of @xmath130 bal is consistent with observation .
note that , @xmath124 in the @xmath13110830 region is the blue curve shown in panel ( a ) of figure [ f4 ] , and the modeled @xmath125 should be added with the black - body emission component ( orange dashed line in figure [ f4 ] ) in the comparison of the model spectrum with observation .
we further explored the various absorption and doublets in the sdss spectrum . here
, the comparison spectrum of sdss j082806.18 + 063608.3 was used to represent the unabsorbed intrinsic spectrum . the observed spectrum ( black curve ) , and modeled spectrum with absorption ( green curve ) for 1523 are plotted in panel ( a ) of figure [ f6 ] .
it can be seen that the modeled spectrum with absorption is in good agreement with the observed one at the red wing of and the longer wavelengths .
the normalized fluxes at wavelength longger than 2800 are around 1 .
that means that the absorption multiplets of opt.6,7 and opt.8 observed in 1523 can be largely recovered using the above modeled absorption . however , the match of opt.6,7 and opt.8 between model and observation is still fluky .
1523 is observed to decrease in flux on the timescale of 8.99 years from the fbqs to the sdss , but the absorption of opt.6,7 and opt.8 changes weakly , and the dramatic absorption variabilities are uv 1 , uv 62 and . a photoionization model with lower density ( @xmath132 ) and ionization parameter ( @xmath133 ) than the above can approximatively match the absorption variations ( see figure 2 of zhang et al .
2015b ) . for the sdss blue - side spectrum
, there seems to be other absorption components for uv 1 , uv 62 and , corresponding to outflow materials with low density .
indeed , the comparison of 1523 and the two templates in panel ( b ) of figure [ f2 ] suggests the existence of the other absorption components . opt.6,7 and opt.8 have similar absorption depth as the balmer absorption troughs , while the troughs of uv 1 are nearly twice as deep as opt.6,7 and opt.8 multiplets .
moreover , the residual fluxes of uv 62 and bal troughs are even only one - third of opt.6,7 and opt.8 troughs .
we broadened a low density model ( @xmath9 , @xmath134 and @xmath129 ) with a single gaussian profile and blueshifted it to approximatively model the residual component shown in panel ( b ) of figure [ f6 ] .
the fwhm of gaussian profile is 3000 km s@xmath4and the blueshifted velocity is 14,000 km s@xmath4 .
if we use a uniform covering factor for this component , @xmath135 is found to be @xmath136 .
we also noticed that the residual absorption spectrum indicate stronger absorption of uv144 - 149 and uv158 - 164 multiplets ( around 2400 ) than the optical model ( panel ( b ) of figure [ f2 ] ) , which most likely suggests a higher density or some microturbulence in the outflow winds ( shi et al . in pre . ) . from the spectral fittings in section 3.1 ,
we can obtain the continuum and emission line parameters , and derive the quasar fundamental parameters .
the measured monochromatic luminosity at 5100 in the rest frame is @xmath137 erg s@xmath15 .
the bolometric luminosity @xmath138 is estimated from the luminosity @xmath139 with a bolometric correction of 9 ( kaspi et al . 2000 ) .
the full width at half - maximum ( fwhm ) of broad balmer emission lines is 1744 km s@xmath4 , and then the bh mass @xmath140 m@xmath141 , is estimated using the following prescription based on the luminosity at 5100 and the width of h@xmath53 emission ( vestergaard & peterson 2006 ) : @xmath142 + ( 6.91 \pm 0.02 ) .
\label{mbh}\ ] ] the corresponding eddington ratio is @xmath143 , @xmath144 .
the black hole mass computing formula of mclure & dunlop ( 2004 ) give the similar results , @xmath145 m@xmath141 and @xmath146 .
1523 is a typical narrow - line seyfert 1 galaxy ( nls1 ) with a low black hole mass but high accretion ratio .
the radius of broad emission line regions ( belrs ) , @xmath147 , can be estimated using the formula based on the luminosity at 5100 , @xmath148 where the parameters , @xmath0 and @xmath53 are @xmath149 and @xmath150 given in greene & ho ( 2005 ) and @xmath151 and @xmath152 given in kaspi et al .
( 2005 ) respectively .
thus , the luminosity yields @xmath153 pc .
the radius of the dust torus , @xmath154 , can also be estimated based on the thermal equilibrium of the inner side of the torus as @xmath155 where @xmath156 is the stefan - boltzmann constant , @xmath157 is the temperature of inner side of the tours .
then , we get @xmath158 pc .
ionization parameter is defined as : @xmath159 in which , @xmath160 is the frequency corresponding the hydrogen edge , and @xmath161 is the ionization photons emission rate . using above ionization equation and the inferred @xmath161 , @xmath126 and @xmath127 values , one can get the distance of the absorption gas of balmer bals from central ionization source , @xmath162 pc .
thus , the high density outflow gas of balmer bals locates at or outside of the belrs , and less than the distance of the dust torus .
similarly , the estimated distance of the low density outflow gas is much farther than the outflow materials with high density , reacheing several tens of parsecs .
ccc cccc ngc 4151 & @xmath163 & @xmath164350 & & & hutchings et al .
( 2002 ) & + sdss j112526.12 + 002901.3@xmath165&@xmath166 , @xmath167&@xmath168 , @xmath169&10@xmath170 & 1.8@xmath171&hall et al .
( 2002)&mntfelobal + sdss j083942.11 + 380526.3@xmath165 & 520 & @xmath164340 & 10@xmath172 & 4 - 40 & aoki et al .
( 2006 ) & + sdss j125942.80 + 121312.6@xmath165 & 3400 & @xmath173 & 10@xmath174 & 1 & hall ( 2007 ) & bal + sdss j102839.11 + 450009.4 & 670 & 149@xmath697 & & & wang et al .
( 2008 ) & + sdss j172341.10 + 555340.5 & 5370 & 450@xmath69130 & & & aoki et al .
( 2010 ) & mntfelobal + lbqs 1206 + 1052 & 726 & 2000 & 10@xmath175 & & ji et al .
( 2012 ) & bal + sdss j222024.59 + 010931.2 & 0 & @xmath1641500 & 10@xmath176 & & ji et al .
( 2013 ) & bal + sdss j112611.63 + 425246.4 & @xmath177&@xmath178 & & & wang & xu ( 2015 ) & + sdss j152350.42 + 391405.2 & 10,353 & @xmath16412,000&10@xmath174 & 0.2 & this paper & bal in 4.1 , we compared balmer bals in 1523 with non - stellar balmer absorption lines in previous literature , and found that 1523 presents the broadest balmer absorption lines with the maximum blueshifted velocity .
then , a question is naturally raised : why does this object present such unique absorption property ? in zhang et al .
( 2014 ) , we know that outflow velocity strongly or moderately depends on the eddington ratio , luminosity , uv and nir slopes ( also see hamann 1998 ; misawa et al .
2007 ; ganguly et al .
the first three items represent the ionization sed and the amount of the high - energy photons , and the last item is the possible contribution of the dust to the outflow acceleration .
1523 is a high - luminous nls1 with near - eddington accretion rate .
meanwhile , the nir slope of 1523 is @xmath179 , which is redder than those of most bal quasars ( see figure 2 in zhang et al .
these statistical properties bring about the occurrence of the high - velocity outflow . from table
[ tab3 ] , we knew that only three of literature quasars are the balmer bals , and others are just the balmer nals .
indeed , the absorption troughs of balmer and @xmath120 series can help us to straighten out the classification of bals .
for example , there are suspected overlapping absorption features of in two objects , hall et al .
( 2002 ) classified them into a special absorption subtype , the so - called many - narrow - trough felobal ( mntfelobal ) .
however , the troughs of balmer and @xmath120 series present the true absorption profiles with narrow widths , and these profiles are used to model absorption multiplets ( shi et al . in pre . ) .
the cloudy simulations provided the physical conditions of outflow gases for part of objects listed in table [ tab3 ] .
it can be seen that almost all absorption lines are constrained to come from high density gases ( @xmath180 ) , and 1523 is among sources with the highest density except sdss j112526.12 + 002901.3 .
the high density suggests that the outflow winds should survive in the inner region of the agn .
the photoionization model gave an estimation of a distance of @xmath181 pc .
the outflow winds in 1523 locate at such close from the central ionizing source which is slightly farther than that of belrs . in the disk wind scenarios ( murray et al .
1995 ) , most photoionized clouds emerge from the accretion disk and accelerate outwards .
the velocities of the clouds are assumed to be a function of the initial velocities , the terminal velocities and the radii that they are away from the center .
the terminal velocities are approximately inversely proportional to the square root of the radius at which the streamlines .
thus the innermost streamlines have the highest rotational and terminal radial velocities and highest ionization states .
balmer absorption winds in 1523 are just located at the distance of @xmath181 pc from the central ionizing source .
we suggest that is the reason why balmer bals in this objetc have very large outflow velocities .
panel ( c ) of figure [ f2 ] shows that the minimum velocity of the bal troughs is @xmath182 km s@xmath4 , that means the absorption winds have large initial velocities when they emerge , or they have been accelerated from ( inside ) the belrs .
if the mass flux is continuous , the optical depths decrease as a function of the blueshifted velocity .
if the outflow is gathering speed , it seems that the volumes of absorption winds expand gradually during acceleration outwards .
that is consistent with the existence of the excess outflow component of uv 62 and bal troughs ( figure [ f6 ] ) .
this component shows higher blueshifted velocity ( @xmath183 km s@xmath4 ) , larger covering factor ( @xmath184 ) and lower density ( @xmath185 ) , and is estimated to exist in remoter regions ( several tens of parsecs ) .
in this work , we present the discovery of balmer - series absorption lines from h@xmath0 , h@xmath53 and h@xmath1 in 1523 from the quasi - simultaneous optical and near - infrared spectroscopy . the redshift of the balmer absorption troughs is @xmath186 , and it is blueshifted by @xmath187 km s@xmath4with regard to the balmer emission lines .
balmer bals have outflowing velocities a factor of two than the previously known balmer absorption lines .
we searched for the same velocity components seen in other nir balmer absorption lines .
we found a component in the @xmath130 absorption line at the same redshift .
the absorption trough in @xmath130 has a uniform absorption profile with the balmer - series , with the absorption width @xmath188 km s@xmath4 .
therefore 1523 is the object with the broadest balmer absorption lines detected so far .
we measured the profiles of balmer bels and derived their widths of @xmath189 km s@xmath4 .
the estimation of the fundamental parameters shows that 1523 is a typical nls1 with a low black hole mass but high accretion ratio .
we approximately evaluate the bals by employing the cloudy in a extensive parameter space . the outflow winds of balmer bals are suggested to be of the electron density of @xmath190 , an ionization parameter of @xmath191 , and the distance of @xmath181 pc from the central ionizing source which is slightly farther than that of belrs .
many thanks to the anonymous referee for the helpful suggestions .
many thanks to dr .
huiyuan wang and dr .
xueguang zhang for very helpful discussions .
this work is supported by chinese natural science foundation ( nsfc-11203021 , 11573024 ) , national basic research program of china ( `` 973 '' program , 2013cb834905 ) the soc program ( chinare-2015 - 02 - 03 ) .
we acknowledge the use of the hale 200-inch telescope at palomar observatory through the telescope access program ( tap ) and the antarctic bright star survey telescope ( bsst ) , as well as the archive data from the first , nvss , sdss , wise and catalina surveys .
the bsst is funded by the `` antarctic inland scientific research station '' program and the soc program ( chinare-2015 - 02 - 03 ) .
tap is funded by the strategic priority research program `` the emergence of cosmological structures '' ( xdb09000000 ) , national astronomical observatories , chinese academy of sciences , and the special fund for astronomy from the ministry of finance .
observations obtained with the hale telescope at palomar observatory were obtained as part of an agreement between the national astronomical observatories , chinese academy of sciences , and the california institute of technology .
the catalina surveys consist of the catalina sky survey ( css ) and the catalina real - time transient survey ( crts ) .
the css survey is funded by the national aeronautics and space administration under grant no .
nng05gf22 g issued through the science mission directorate near - earth objects observations program .
the crts survey is supported by the u.s .
national science foundation under grants ast-0909182 .
funding for sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , and the u.s .
department of energy office of science .
the sdss - iii web site is http://www.sdss3.org/. adelman - mccarthy , j. k. , ageros , m. a. , allam , s. s. , et al .
2007 , , 172 , 634 ai , y. l. , yuan , w. , zhou , h. y. , et al .
2010 , , 716 , l31 antonuccio - delogu , v. , & silk , j. 2010 , accretion and ejection in agn : a global view , 427 , 343 aoki , k. , iwata , i. , ohta , k. , et al .
2006 , , 651 , 84 aoki , k. 2010 , , 62 , 1333 arav , n. , de kool , m. , korista , k. t. , et al .
2001 , , 561 , 118 baskin , a. , laor , a. , & hamann , f. 2013 , , 432 , 1525 becker , r. h. , white , r. l. , & helfand , d. j. 1995 , , 450 , 559 becker , r. h. , white , r. l. , gregg , m. d. , et al .
2000 , , 538 , 72 brandt , w. n. , laor , a. , & wills , b. j. 2000 , , 528 , 637 brinkmann , w. , wang , t. , matsuoka , m. , & yuan , w. 1999 , , 345 , 43 condon , j. j. , cotton , w. d. , greisen , e. w. , et al .
1998 , , 115 , 1693 dong , x. , wang , t. , wang , j. , et al .
2008 , , 383 , 581 drake , a. j. , et al .
2014 , , 213 , 9 fan , l. l. , wang , h. y. , wang , t. , et al .
2009 , , 690 , 1006 ferland , g. j. , korista , k. t. , verner , d. a. , et al .
1998 , , 110 , 761 forster , k. , green , p. j. , aldcroft , t. l. , et al .
2001 , , 134 , 35 gallagher , s. c. , brandt , w. n. , chartas , g. , & garmire , g. p. 2002
, , 567 , 37 gallagher , s. c. , brandt , w. n. , chartas , g. , et al .
2006 , , 644 , 709 ganguly , r. , brotherton , m. s. , cales , s. , et al .
2007 , , 665 , 990 gibson , r. r. , brandt , w. n. , schneider , d. p. , & gallagher , s. c. 2008 , , 675 , 985 gibson , r. r. , jiang , l. , brandt , w. n. , et al .
2009 , , 692 , 758 gibson , r. r. , brandt , w. n. , gallagher , s. c. , et al .
2010 , , 713 , 220 granato , g. l. , de zotti , g. , silva , l. , bressan , a. , & danese , l. 2004 , , 600 , 580 green , p. j. , schartel , n. , anderson , s. f. , et al .
1995 , , 450 , 51 greene , j. e. , & ho , l. c. 2005 , , 630 , 122 hall , p. b. 2007 , , 133 , 1271 hall , p. b. , anderson , s. f. , strauss , m. a. , et al .
2002 , , 141 , 267 hall , p. b. , hutsemkers , d. , anderson , s. f. , et al .
2003 , , 593 , 189 hall , p. b. , anosov , k. , white , r. l. , et al .
2011 , , 411 , 2653 hamann , f. 1998 , , 500 , 798 hopkins , p. f. , hernquist , l. , cox , t. j. , & kere , d. 2008 , , 175 , 356 hutchings , j. b. , crenshaw , d. m. , kraemer , s. b. , et al .
2002 , , 124 , 2543 ji , t. , wang , t .-
, zhou , h .- y . , &
wang , h .- y .
2012 , research in astronomy and astrophysics , 12 , 369 ji , t. , zhou , h .- y . ,
wang , t .-
, & wang , h .- y . 2013 , cha&a , 37 , 17 kaspi , s. , smith , p. s. ,
netzer , h. , et al .
2000 , , 533 , 631 kaspi , s. , maoz , d. , netzer , h. , et al .
2005 , , 629 , 61 kraemer , s. b. , crenshaw , d. m. , hutchings , j. b. , et al .
2001 , , 551 , 671 krongold , y. , binette , l. , & hernndez - ibarra , f. 2010 , , 724 , l203 landolt , a. u. 2009 , , 137 , 4186 landt , h. , bentz , m. c. , ward , m. j. , et al .
2008 , , 174 , 282 laor , a. , & brandt , w. n. 2002 , , 569 , 641 mclure , r. j. , & dunlop , j. s. 2004 , , 352 , 1390 liu , w .- j .
, zhou , h. , ji , t. , et al .
2015 , , 217 , 11 misawa , t. , charlton , j. c. , eracleous , m. , et al .
2007 , , 171 , 1 murray , n. , chiang , j. , grossman , s. a. , & voit , g. m. 1995 , , 451 , 498 reichard , t. a. , richards , g. t. , hall , p. b. , et al .
2003 , , 126 , 2594 scannapieco , e. , & oh , s. p. 2004
, , 608 , 62 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 skrutskie , m. f. , cutri , r. m. , stiening , r. , et al .
2006 , , 131 , 1163 stoughton , c. , lupton , r. h. , bernardi , m. , et al .
2002 , , 123 , 485 trump , j. r. , hall , p. b. , reichard , t. a. , et al .
2006 , , 165 , 1 vacca , w. d. , cushing , m. c. , & rayner , j. t. 2003 , , 115 , 389 vron - cetty , m .-
p . , joly , m. , & vron , p. 2004
, , 417 , 515 vron - cetty , m .-
p . , joly , m. , vron , p. , et al .
2006 , , 451 , 851 vestergaard , m. , & peterson , b. m. 2006 , , 641 , 689 wang , t. g. , wang , j. x. , brinkmann , w. , & matsuoka , m. 1999 , , 519 , l35 wang , t. , dai , h. , & zhou , h. 2008 , , 674 , 668 wang , h. , xing , f. , zhang , k. , et al .
2013 , , 776 , l15 wang , j. , & xu , d. w. 2015 , , 573 , aa15 weymann , r. j. , morris , s. l. , foltz , c. b. , & hewett , p. c. 1991 , , 373 , 23 wilson , j. c. , henderson , c. p. , herter , t. l. , et al . 2004 , , 5492 , 1295 wright , e. l. , eisenhardt , p. r. m. , mainzer , a. k. , et al . 2010 , , 140 , 1868 zhang , s. , wang , t .-
g . , wang , h. , et al .
2010 , , 714 , 367 zhang , s .- h . ,
wang , h .- y .
, zhou , h .- y . ,
wang , t .-
, & jiang , p. 2011
, research in astronomy and astrophysics , 11 , 1163 zhang , s. , wang , h. , wang , t. , et al .
2014 , , 786 , 42 zhang , s. , ge , j. , jiang , p. , et al . 2015a , , 802 , 92 zhang , s. , zhou , h. , wang , t. , et al . 2015b , , 803 , 58 zhou , h. , wang , t. , wang , h. , wang , j. , yuan , w. , & lu , y. 2006 , , 639 , 716 | we present the discovery of balmer line absorption from h@xmath0 to h@xmath1 in an iron low - ionization broad absorption line ( felobal ) quasar sdss j152350.42 + 391405.2 ( hereafter 1523 ) , by the quasi - simultaneous optical and near - infrared spectroscopy .
the balmer line absorption is at @xmath2 and blueshifted by @xmath3 km s@xmath4with respect to the balmer emission lines .
all balmer bals have uniform absorption profile with the widths of @xmath512,000 km s@xmath4 .
we also found the absorption trough in * @xmath610830 with the same velocity and width in the @xmath7-band triplespec spectrum of 1523 .
this object is only the tenth active galactic nucleus known to exhibit non - stellar balmer absorption , and also the case with the highest velocity and broadest balmer absorption lines which have ever been found .
a cloudy analysis shows that the absorbers require an gas density of @xmath8 and an ionization parameter of @xmath9 .
they locate at a distance of @xmath10 pc from the central ionizing source which is slightly farther than that of belrs . furthermore
, 1523 is one of the brightest balmer - bal quasar ever reported , with unique iron absorption variations , making it as the most promising candidate for follow up high - resolution spectroscopy , multi - band observations , and long - term monitoring . |
the search for a higgs boson is the primary goal of all present and future high energy experiments at the tevatron , lhc or an @xmath5 linear collider . whereas the standard model ( sm ) predicts just one higgs boson , with the present lower bound of its mass @xmath6 ( at 95% confidence level ) @xcite , extensions of the sm allow for more higgs bosons . in particular , the minimal supersymmetric standard model ( mssm ) contains five physical higgs bosons : two neutral cp even ( @xmath7 and @xmath8 ) , one neutral cp odd ( @xmath0 ) , and two charged ones ( @xmath9 ) @xcite .
the existence of a charged higgs boson or a cp odd neutral one would give clear evidence for physics beyond the sm .
for a discovery precise predictions for its decay modes and their branching ratios are necessary . in case
supersymmetric ( susy ) particles are not too heavy , the higgs bosons can also decay into susy particles ( neutralinos @xmath10 , charginos @xmath11 , sfermions @xmath12 ) , @xmath13 ( @xmath14 = 1 4 ) , @xmath15 , @xmath16 ( @xmath17 = 1 , 2 ) , @xmath18 , @xmath19 . at tree
level , these decays were studied in @xcite . in particular ,
the branching ratios for the decays into sfermions , @xmath16 , can be sizeable depending on the parameter space @xcite . the susy
qcd corrections to the decays into sfermions have also been calculated @xcite .
the corrections to the decays into neutralinos @xcite or charginos @xcite due to fermion / sfermion exchanges have also been found non
+ in this paper , we study in detail the decay of the cp
odd higgs boson @xmath0 into two sfermions , @xmath20 .
in particular , the third generation sfermions @xmath21 , and @xmath22 are interesting because one expects them to be lighter than the other sfermions due to their large yukawa couplings and left
right mixings . since @xmath0 only couples to @xmath23@xmath24 ( left right states of @xmath25 ) , and due to the cp nature of @xmath0 , @xmath26 vanishes .
( this is valid also beyond tree
level for real parameters in the mssm . )
we will calculate the _ full _ electroweak corrections in the on shell scheme .
owing to the fact that almost all parameters of the mssm have to be renormalized in this process and hence a large number of graphs has to be computed , the calculation is very complex . despite this complexity
, we have performed the calculation in an analytic way .
we have also studied the crossed channel @xmath27 .
some important numerical results , especially for the decays of @xmath0 into the third generation squarks and the corresponding crossed channels have already been shown in @xcite .
+ this paper has three new elements .
first , we give in the appendix all analytical formulae needed for the calculation .
second , we describe in detail our method for improving the full one loop calculation . as pointed out in @xcite , in the case of the decay of @xmath28 or @xmath29 the decay widths can receive large corrections , especially for large @xmath1 .
this makes the perturbation expansion unreliable . in some cases the width can even become negative in the on shell renormalization scheme .
we present here a detailed description how this problem can be solved by an appropriate definition of the tree
level coupling in terms of running fermion masses and running trilinear coupling @xmath2 ( @xmath30 ) . for consistency between the on shell and the @xmath31 running parameters an iteration procedure
is necessary .
moreover , in addition to the numerical results shown in @xcite we present here new results for @xmath32 , @xmath33 , @xmath34 and the corresponding crossed channels . + the paper is organized as follows . in section [ treelevel ]
we summarize the tree level formulae . in section [ fecorr ]
the full electroweak corrections are presented for which the explicit analytic formulae are given in the appendices [ appvertex ] , [ appwf ] , [ appse ] . in section [ imp ] a detailed description of the method to improve the one loop calculation is presented . in section [ method ] the numerical treatment as well as numerical results are shown .
section [ conclusions ] summarizes our conclusions .
the sfermion mixing is described by the sfermion mass matrix in the left
right basis @xmath35 , and in the mass basis @xmath36 , @xmath37 or @xmath38 , @xmath39 a_f\,m_f & m_{\sf_r}^{\,2 } \end{array } \right ) = \left ( r^\sf \right)^\dag \left ( \begin{array}{cc } m_{\sf_1}^{\,2 } & 0 \\[2 mm ] 0 & m_{\sf_2}^{\,2 } \end{array } \right ) r^\sf \,,\end{aligned}\ ] ] where @xmath40 is a 2 x 2 rotation matrix with rotation angle @xmath41 , which relates the mass eigenstates @xmath42 , @xmath43 , @xmath44 to the gauge eigenstates @xmath45 , @xmath46 , by @xmath47 and @xmath48\label{msd } m_{\sf_r}^{\,2 } & = & m_{\{\ti u\!,\,\ti d\!,\,\ti e \}}^2 + e_{f}\sin^2\!\tw \cos2\b\,m_{\scriptscriptstyle z}^{\,2 } + m_f^2\ , , \\[2 mm ] a_f & = & a_f - \mu \,(\tan\b)^{-2 i^{3l}_f } \,.\end{aligned}\ ] ] @xmath49 , @xmath50 , @xmath51 , @xmath52 and @xmath53 are soft susy breaking masses , @xmath2 is the trilinear scalar coupling parameter , @xmath54 the higgsino mass parameter , @xmath55 is the ratio of the vacuum expectation values of the two neutral higgs doublet states @xcite , @xmath56 denotes the third component of the weak isospin of the fermion @xmath57 , @xmath58 the electric charge in terms of the elementary charge @xmath59 , and @xmath60 is the weinberg angle .
+ the mass eigenvalues and the mixing angle in terms of primary parameters are @xmath61 and the trilinear breaking parameter @xmath2 can be written as @xmath62 with @xmath63 .
+ at tree
level the decay width of @xmath64 is given by @xmath65 and the colour factor @xmath66 for squarks and for sleptons , respectively .
@xmath67 denotes the @xmath0@xmath68@xmath69 coupling as given in appendix [ appcouplings ] .
the full one loop corrected decay width is given by @xmath70 \,.\end{aligned}\ ] ] the ( uv finite ) corrections @xmath71 consist of the vertex corrections @xmath72 ( fig .
[ vertex - graphs ] ) , wave function corrections and the coupling counter term corrections @xmath73 owing to the shift from the bare to the on shell values , @xmath74 the renormalization procedure with the fixings of the counter terms is given in @xcite .
the explicit formulae of the vertex corrections @xmath72 as well as the various contributions to the wave function corrections @xmath75 , @xmath76 g_{123}^\sf \ , , \end{aligned}\ ] ] can be found in the appendices [ appvertex ] and [ appwf ] .
for the explicit formulae of the self
energies needed for the calculation of the counter term correction , @xmath73 ( see eq . ( 23 ) of @xcite ) , we refer to the appendix [ appse ] .
+ due to the diagrams with photon exchange we also have to consider real photon emission corrections to cancel the infrared divergences ( fig .
[ vertex - graphs ] ) . therefore the corrected ( uv and ir convergent ) decay width is @xmath77 throughout the paper we use the susy invariant dimensional reduction @xmath78 as regularization scheme . for convenience
we perform the calculation in the t hooft feynman gauge , @xmath79 .
it has already pointed out in @xcite that in the case of bottom squarks and tau sleptons , especially for large @xmath1 the corrections to the decay widths @xmath80 and @xmath29 can be very large in the on shell renormalization scheme .
if the corrections are negative , the one loop corrected width can even become negative and therefore unphysical . hence the perturbation expansion around the on shell tree
level is no longer reliable .
it has been shown in @xcite that , in the case of the decays into bottom squarks , the source of these large corrections are mainly the counter terms for @xmath81 and the trilinear coupling @xmath82 , in particular the susy qcd corrections . however , despite the absence of strong interactions for the decay into tau sleptons , the corrections become extremely large . we show that this problem can be solved by defining an appropriate tree
level in terms of running values for @xmath83 and @xmath2 .
the expansion around this new tree level then no longer suffers from bad convergence . +
* correction to [ sec : mbmssm ] * + first we review the improvement of the perturbation expansion by using @xmath31 running bottom quark masses , following @xcite . if the yukawa coupling @xmath84 is given at tree
level in terms of the pole mass @xmath81 , the one loop corrections become very large due to gluon and gluino exchange contributions to the counter term @xmath85 . the large counter term caused by the gluon loop
is absorbed by using sm 2loop renormalization group equations in the @xmath86 scheme @xcite .
thus we obtain the sm running bottom @xmath87 : @xmath88 the ratio @xmath89 can be expressed as @xmath90 { \displaystyle { \frac{c_6(\alpha_s^{(2)}(q)/\pi)}{c_6(\alpha_s^{(2)}(m_t)/\pi)}\ , \frac{c_5(\alpha_s^{(2)}(m_t)/\pi)}{c_5(\alpha_s^{(2)}(m_b)/\pi ) } } } & ( q >
m_t ) \ , , \end{array } \right.\end{aligned}\ ] ] where we have used the functions @xmath91 { \displaystyle { c_6(x ) ~=~ \left(\frac{7}{2}x\right)^{\frac{4}{7}}(1 + 1.398x ) } } & ( q > m_t)\ , , \end{array}\end{aligned}\ ] ] and the 2loop rges for @xmath92 @xcite , @xmath93 with @xmath94 or 6 for @xmath95 or @xmath96 , respectively . for the sm @xmath31
running bottom mass at the scale @xmath97 we use the @xmath86 equation @xmath98^{-1}\ , , \ ] ] with @xmath99 and then convert to @xmath31 using one loop running @xmath100 : @xmath101 in the mssm , for large @xmath1 the counter term to @xmath81 can be very large due to the gluino mediated graph @xcite . here
we absorb the gluino contribution as well as the sizeable contributions from neutralino and chargino loops and the remaining electroweak self
energies into the higgs sfermion sfermion tree
level coupling .
in such a way we obtain the full @xmath31 running bottom quark mass @xmath102 the explicit form of the electroweak contribution to the counter term @xmath103 is given in appendix [ appfermion - se ] . + * correction to * + the second source of a very large correction ( in the on shell scheme ) are the counter terms for the trilinear coupling @xmath104 , ( see eq .
[ mfaf ] ) , @xmath105 again , the big bottom mass correction @xmath85 contributes to @xmath106 , but also the counter term of the left
right mixing elements of the sfermion mass matrix , @xmath107 , gives a very large correction for higher values of @xmath1 .
in particular , in the case of the decay into staus , this is the main source for the bad convergence of the tree
level expansion . as in the case of the large correction to @xmath81
we redefine the higgs
sfermion sfermion tree
level coupling in terms of @xmath31 running @xmath108 . because of the fact that the counter terms @xmath109 ( for large @xmath1 ) can become several orders of magnitude larger than the on
shell @xmath104 we use @xmath108 as input @xcite . in order to be consistent
we have to perform an iteration procedure to get all the correct running and on shell masses , mixing angles and other parameters
. this procedure is described below .
in this section we will explain in detail how we can improve the perturbation calculation for the sbottom and stau case by using @xmath31 running values for @xmath81 and @xmath104 in the higgs
sfermion sfermion tree level couplings .
since we take @xmath31 running values for @xmath110 and @xmath111 as input and all other parameters on shell we will have to pay attention to the sbottom and stau sector in order to get consistently all needed running and on shell masses , mixing angles and other parameters .
here we adopt the procedure developed in @xcite and also extend it to the electroweak case .
* stop sector : * + we start our calculation in the stop sector . because all input parameters in the stop sector are on shell
we obtain the on shell masses @xmath112 , @xmath113 and the stop mixing angle @xmath114 by diagonalizing the stop mass matrix in the @xmath115@xmath116 basis , see chapter [ treelevel ] .
the running stop masses @xmath117 and mixing angle @xmath118 are calculated at the scale @xmath119 by adding the appropriate counter terms to the on shell values in @xmath120 \hat \theta_\st ( q_\st ) & = & \theta_\st + { \delta}\theta_\st\,.\end{aligned}\ ] ] the electroweak parts of the sfermion self
energies @xmath121 are given in appendix [ appsfermion - se ] and the susy
qcd contributions , @xmath122 are given in eqs .
( 25)(27 ) in @xcite .
here and in the following all running parameters @xmath123 are related to their on shell values @xmath124 by @xmath125 , with @xmath126 being the full one loop counter term also including the susy qcd parts .
the mixing angle is fixed by @xcite @xmath127 for @xmath31 running @xmath128 we use the formulae from section [ sec : mbmssm ] with the obvious substitutions @xmath129 and @xmath130 for the top case .
next we evaluate the running parameters @xmath131 and @xmath132 by inserting the running values @xmath133 , @xmath134 , @xmath135 and @xmath136 into the equations @xmath137
m_{\ti u}^{2 } & = & m_{\st_1}^2 \sin^{2}\theta_{\st } + m_{\st_2}^2 \cos^{2}\theta_{\st } - m_{t}^{2 } - m_{\scriptscriptstyle z}^2\cos 2\beta \,e_t\sin^2 \tw \ , .\end{aligned}\ ] ] for the running value of @xmath138 we use ( see eq .
( [ mfaf ] ) ) @xmath139 where we have taken running @xmath140 @xcite . + * sbottom sector : * + in the sbottom sector we have given all parameters on shell except the parameter for the trilinear coupling , @xmath141 , which is running .
first we calculate @xmath142 from eq .
( [ mbmssm ] ) at the scale @xmath143 . from the stop sector
we already know the running values of @xmath144 and @xmath54
. then we diagonalize the sbottom mass matrix using @xmath145 and on shell @xmath52 , which is near its running value @xmath146 , to obtain the starting values for @xmath147 and @xmath148 .
the on shell sbottom masses @xmath149 and the mixing angle @xmath150 are calculated from their running values by subtracting the appropriate counter terms , i. e. @xmath151 , @xmath152 .
now we can compute the running value for @xmath52 . using the relation (
( [ msd ] ) ) @xmath153 we get @xmath154 with @xmath155 and @xmath156 .
because the parameters involved in these calculations are very entangled , e. g. @xmath146 depends on the @xmath157 which themselves depend on @xmath52 , we have to perform an iteration procedure . + * iteration procedure : * + here we will describe in detail the procedure how we obtain all necessary on shell and running parameters . for convenience , we shortly denote all masses , parameters , couplings etc . for a certain @xmath158 in the iteration by @xmath159 . as starting values @xmath160 we take on shell masses and parameters ( except @xmath110 which is running ) and the couplings derived from these quantities .
the only exceptions are the standard model running fermion masses @xmath161 .
@xmath162 shortly stands for the full @xmath31 running fermion masses , @xmath163 . +
the single steps of the iteration procedure are the following : 1 .
the running stop masses and the stop mixing angle are calculated as explained + above by @xmath164 and @xmath165 .
@xmath166 3 .
@xmath167 and + @xmath168 with @xmath169 4 . the running value of @xmath1 , @xmath170 , + with @xmath171 @xcite .
5 . @xmath172 with @xmath173
the soft susy breaking masses @xmath174 are calculated from @xmath175 , @xmath176 , @xmath177 , @xmath178 , @xmath179 and @xmath180 . 7 .
we compute the running @xmath181 by using running values in eq .
( [ atr ] ) : + @xmath182 8 .
[ sbstepstart ] in the sbottom sector we obtain @xmath183 from the running values already calculated in steps 1.7 . , like @xmath184 , @xmath185 or @xmath186 , and the remaining masses , couplings etc . from @xmath187
@xmath188 . 10 .
we receive the running sbottom masses , @xmath189 , and the mixing angle , @xmath185 , by solving the mass eigenvalue problem with the running values of @xmath190 , @xmath191 , @xmath192 , @xmath110 , @xmath193 and @xmath194 .
the on shell sbottom masses @xmath195 at the scale @xmath196 , and @xmath197 . 12 .
@xmath198 + @xmath199 .
+ ( remember that the values without a hat ( @xmath200 ) are on shell ones ! ) 13 .
[ sbstepend ] @xmath201 . 14 . in the sneutrino sector
we calculate the running sneutrino mass + @xmath202 and @xmath203 , see also eq ( [ msq2 ] ) . 15 . in the stau sector
the values for running @xmath204 etc . are calculated like in the steps [ sbstepstart][sbstepend ] in the sbottom sector with the evident substitution @xmath205 for the corresponding parameters and @xmath206 .
all couplings are recalculated with the new running parameters @xmath207 .
the iteration starts with @xmath208 and ends , when certain parameters are calculated precisely enough for a given accuracy , i. e. @xmath209 for @xmath210 .
for @xmath211 we choose @xmath212 .
we have checked the consistency of this procedure by computing the on
shell @xmath52 and running @xmath49 from the sbottom sector by using @xmath213 which are equal ( up to higher order corrections ) to the on shell input @xmath52 and running @xmath49 from the stop sector .
+ for easier reading the single steps of the iteration procedure of the stop and sbottom sector are depicted in the flowchart in fig .
[ flowchart ] .
( 160,150)(0,0 ) ( 10,5 ) in the following numerical examples , we take for the standard model parameters @xmath214 gev , @xmath215 gev , @xmath216 , @xmath217 , @xmath218 gev , @xmath219 gev , @xmath220 gev and @xmath221 @xmath222 mev for @xmath223 and @xmath224 generation fermions .
@xmath225 is fixed by the gaugino unification relation @xmath226 , therefore the gluino mass is related to @xmath227 by @xmath228 . in order to reduce the number of parameters in the input parameter set , we assume @xmath229 for the first , second and third generation soft susy breaking masses as well as @xmath230 for all ( s)fermion generations , if not stated otherwise .
+ * stop case : * in fig .
[ mudep_st_os ] we show the tree level and the corrected widths to @xmath231 for @xmath232 and @xmath233 as a function of the higgsino mass parameter @xmath54 .
the electroweak corrections are almost constant about @xmath234% . + at @xmath235 one can identify the pseudo threshold coming from @xmath236 .
( 160,65)(0,0 ) ( 30,5 ) ( 74,3)(0,0)[t]@xmath54 [ gev ] ( 20,12 ) fig .
[ mst1dep_st_os ] shows the tree
level , the full electroweak and the full one loop corrected ( electroweak and susy qcd ) decay width of @xmath231 as a function of the lighter stop mass , @xmath112 , where @xmath237 is varied from 200 to 450 gev . as input parameters we choose @xmath238 and @xmath239 . again
, in a large region of the parameter space the electroweak corrections are comparable to the susy qcd ones . the pseudo threshold at @xmath240 originates from @xmath241 in the wave function correction .
( 160,65)(0,0 ) ( 29,5 ) ( 76,3)(0,0)[t]@xmath112 [ gev ] ( 20,12 ) in fig .
[ tanbdep_st_os_crossed ] the dependence of the crossed channel decay width , @xmath242 , as a function of @xmath1 is given .
we see that the electroweak corrections have different sign compared to the susy qcd ones and go up to 10% . as input parameters we have chosen @xmath243 as well as @xmath244 to get an acceptable stop mass splitting .
( 160,65)(0,0 ) ( 28.5,5 ) ( 76,3)(0,0)[t]@xmath1 ( 20,20 ) fig .
[ mst1dep_st_os_crossed ] shows the decay width @xmath245 as a function of @xmath112 , varying @xmath246 from 200 to 460 gev . to get a larger mass splitting for the top squarks , we relax the conditions for 3rd generation squarks and take @xmath247 .
all other susy breaking masses are fixed at 300 gev . for the remaining input parameters we choose @xmath248 and @xmath249 .
( 160,65)(0,0 ) ( 28.5,5 ) ( 76,3)(0,0)[t]@xmath112 [ gev ] ( 20,20 ) * sbottom case : * in fig .
[ tanbdep_sb_imp ] we show two kinds of perturbation expansion for @xmath250 with @xmath251 @xmath252 : first we show the tree
level width , given in terms of on shell input parameters ( dotted line ) .
the dashed and dash dot dotted line correspond to the on shell electroweak and full ( electroweak plus susy qcd ) one loop width , respectively . for both corrections
one can clearly see the invalidity of the on shell perturbation expansion , which leads to an improper negative decay width .
the second way of perturbation expansion is given by the dash dotted and the solid line ( 160,65)(0,0 ) ( 26,5 ) ( 76,3)(0,0)[t]@xmath1 ( 20,12 ) which correspond to the improved tree level and improved full one loop decay width , respectively . here
we take the same input parameters as in the first case but with running @xmath253 . in fig .
[ adep_sb_imp ] we show the decay width @xmath254 as a function of @xmath31 running @xmath255 for @xmath256 and @xmath257 .
the dotted line corresponds to the improved tree
level , the dashed line corresponds to the improved susy
qcd one loop width and the solid line shows the full improved one loop width . for negative @xmath255
the electroweak corrections decrease the decay width by @xmath258 , whereas for positive @xmath255 the susy qcd corrections almost vanish and the electroweak ones go up to @xmath259 .
( 160,65)(0,0 ) ( 31,5 ) ( 76,3)(0,0)[t]@xmath255 [ gev ] ( 20,11 ) ( 160,58)(0,0 ) ( 28.5,5 ) ( 76,3)(0,0)[t]@xmath1 ( 20,20 ) fig .
[ tanbdep_sb_imp_crossed ] shows the behaviour of the decay width @xmath260 for large @xmath1 .
as in fig .
[ tanbdep_sb_imp ] two kinds of perturbation expansion are given .
the dotted and dash dot dotted lines correspond to the tree level and full one loop decay widths in the pure on shell scheme . for
large @xmath1 one can clearly see the invalidity of the perturbation series , leading to a negative decay width . in the second case
we show the expansion around the tree
level decay width , given in terms of running @xmath82 and @xmath81 .
the dash dotted line corresponds to the improved tree level and the solid one to the one loop width .
up to @xmath261 the corrections stay relatively small which indicates that already the ( improved ) tree
level is a good approximation for @xmath260 . as input parameters we take the values @xmath262 and @xmath263 for kinematical reasons . in fig .
[ tanbdep_stau_imp ] the @xmath0 decay into two staus is given as a function of @xmath1 . despite the absence of susy qcd corrections the perturbation expansion around the on shell tree level ( dotted line ) leads to an improper negative decay width ( dashed line ) coming from large @xmath264 corrections .
as input parameters we take @xmath265 .
the dash dotted line corresponds to the improved tree
level and the solid line shows the improved one loop width for the same input parameters as above and running @xmath266 .
( 160,65)(0,0 ) ( 28,5 ) ( 76,3)(0,0)[t]@xmath1 ( 20,12 ) fig .
[ madep_stau_imp ] shows the decay width of @xmath267 as a function of the mass of the decaying higgs boson @xmath0 for the improved perturbation expansion .
the dotted and the solid lines correspond to the ( improved ) tree level and full one loop widths , respectively . in the whole region of the parameter space shown the ( electroweak ) corrections decrease the on shell width by 15% . as input parameters we choose @xmath268 and @xmath239 .
( 160,65)(0,0 ) ( 29,5 ) ( 76,3)(0,0)[t]@xmath269 ( 20,12 ) in fig .
[ adep_stau_imp_crossed ] we show the @xmath255 dependence of @xmath270 in the improved case . for negative values of @xmath255 the corrections increase the on shell width by @xmath271 whereas for positive values of @xmath255
the corrections are negative and go up to @xmath272 .
the input parameters are taken as follows : @xmath273 and @xmath274 for an acceptable stau mass splitting .
for @xmath1 we take the value 30 .
( 160,65)(0,0 ) ( 28.5,5 ) ( 76,3)(0,0)[t]@xmath255 ( 20,20 )
we have calculated the _ full _ electroweak one loop corrections to the decay widths @xmath275 and @xmath4 in the on shell scheme .
we have presented all formulae required for the computation .
it has been necessary to renormalize almost all parameters of the mssm .
we have also included the susy
qcd corrections which were calculated in @xcite . for the decay into bottom squarks and tau sleptons for large @xmath1 an improvement of the on
shell perturbation expansion is necessary .
we have worked out an iterative method to improve the one loop calculation .
thereby , the tree
level coupling is redefined in terms of @xmath31 running masses and running @xmath2 .
we find that the corrections are significant and in a wide range of the parameter space comparable to the susy
qcd corrections .
+ * acknowledgements * + the authors acknowledge support from eu under the hprn - ct-2000 - 00149 network programme and the `` fonds zur frderung der wissenschaftlichen forschung '' of austria , project no . p13139-phy .
for the neutral and charged higgs fields we use the notation @xmath276 , @xmath277 and @xmath278 .
@xmath279 stands for an up type ( s)fermion and @xmath280 for a down type one .
following @xcite the higgs
sfermion couplings for neutral higgs bosons , @xmath281 , can be written as @xmath282_{ij } \,.\end{aligned}\ ] ] the 3rd generation left
right couplings @xmath283 for up and down type sfermions are @xmath284 \non g_{lr , 1}^\sb & = & \left(\ !
\begin{array}{cc } \sqrt 2 h_b m_b s_\a + g_{\scriptscriptstyle z } m_{\scriptscriptstyle z } ( i_b^{3l}\!-\!e_b s_{\scriptscriptstyle w}^2 ) s_{\a+\b } & \frac{h_b}{\sqrt 2 } ( a_b\ , s_\a + \mu c_\a ) \\ \frac{h_b}{\sqrt 2 } ( a_b\ , s_\a + \mu c_\a ) & \sqrt 2 h_b m_b s_\a + g_{\scriptscriptstyle z } m_{\scriptscriptstyle z } e_b s_{\scriptscriptstyle w}^2 s_{\a+\b } \end{array } \!\right ) \ , , \\[2 mm ] \non & & g_{lr , 2}^\sf ~=~ g_{lr , 1}^\sf \qquad { \rm with}\ \a \rightarrow \a - \pi/2 \ , , \\[2 mm ] \non g_{lr , 3}^\st & = & -\sqrt 2 h_t \left(\
! \begin{array}{cc } 0 & -\frac{i}{2 } \big ( a_t\ , c_\b + \mu\ , s_\b \big ) \\
\frac{i}{2 } \big ( a_t\ , c_\b + \mu\ , s_\b \big ) & 0 \end{array } \!\right ) \ , , \\[2 mm ] \non g_{lr , 3}^\sb & = & -\sqrt 2 h_b \left(\ !
\begin{array}{cc } 0 & -\frac{i}{2 } \big ( a_b\ , s_\b + \mu\ , c_\b \big ) \\
\frac{i}{2 } \big ( a_b\ , s_\b + \mu\ , c_\b
\big ) & 0 \end{array } \!\right ) \ , , \\[2 mm ] \non & & g_{lr , 4}^\sf ~=~ g_{lr , 3}^\sf \qquad { \rm with}\ \b \rightarrow \b - \pi/2 \,,\end{aligned}\ ] ] where we have used the abbreviations @xmath285 , @xmath286 and @xmath287 .
@xmath288 denotes the mixing angle of the @xmath289system , and @xmath290 and @xmath84 are the yukawa couplings @xmath291 the couplings of charged higgs bosons to two sfermions are given by @xmath292 @xmath293 @xmath294 h_t ( a_t \cos\b + \mu \sin\b ) & h_t m_b \cos\b + h_b m_t \sin\b \end{array } \right ) \ , , \\[4mm]\non g_{lr , 1}^{\sb\st } & = & \left ( \begin{array}{cc } h_b m_b \sin\b + h_t m_t \cos\b - \frac{g m_{\scriptscriptstyle w}}{\sqrt 2 } \sin 2\b & h_t ( a_t \cos\b + \mu \sin\b ) \\[2 mm ] h_b ( a_b \sin\b + \mu \cos\b ) & h_t m_b \cos\b + h_b m_t \sin\b \end{array } \right ) ~=~ \left ( g_{lr , 1}^{\st\sb } \right)^t \ , , \\ \\
g_{lr , 2}^{\sf\sf ' } & = & g_{lr , 1}^{\sf\sf ' } \qquad { \textrm{with } } \quad \b ~\rightarrow~ \b - \frac{\pi}{2 } \,.\end{aligned}\ ] ] @xmath295 denotes the isospin partner of the fermion @xmath57 , i. e. @xmath296 etc .
note that only the angle @xmath297 explicitly given in the matrices above has to be substituted ; the dependence of @xmath297 in the yukawa couplings has to remain the same .
+ the @xmath298 interaction is given by @xmath299\ , h^0_k h^0_l \sf_i^ * \sf_j \,,\ ] ] with @xmath300 c_{kl}^\st & = & \left ( \begin{array}{cccc } \cos^2\!\alpha & \frac{1}{2 } \sin2\alpha & 0 & 0 \\
\frac{1}{2 } \sin2\alpha & \sin^2\!\alpha & 0 & 0 \\ 0 & 0 & \cos^2\!\beta & \frac{1}{2 } \sin2\beta \\ 0 & 0 & \frac{1}{2 } \sin2\beta & \sin^2\!\beta \end{array } \right)\ , , \\[3 mm ] e_{ij}^\sf & = & \frac{1}{2 c_{\scriptscriptstyle w}^2 } \bigg [ ( i_f^{3l } - e_f s_{\scriptscriptstyle w}^2 ) r_{i1}^\sf r_{j1}^\sf + e_f s_{\scriptscriptstyle w}^2 r_{i2}^\sf r_{j2}^\sf \bigg ] \ , .
\end{aligned}\ ] ] for the @xmath301 interaction , @xmath302\ , h^+_k h^-_l
\sf_i^ * \sf_j \,,\end{aligned}\ ] ] we use the coupling matrices @xmath303 d_{kl}^\st & = & \left ( \begin{array}{cccc } \cos^2\!\beta & \frac{1}{2 } \sin2\beta & 0 & 0 \\ \frac{1}{2 } \sin2\beta & \sin^2\!\beta & 0 & 0 \\ 0 & 0 & \cos^2\!\beta & \frac{1}{2 } \sin2\beta \\ 0 & 0 & \frac{1}{2 } \sin2\beta & \sin^2\!\beta \end{array } \right)\ , , \\[3 mm ] f^\sf_{ij } & = & \frac{1}{2 c_{\scriptscriptstyle w}^2 } \bigg [ \big ( - i_f^{3l } \cos 2\theta_{\scriptscriptstyle w } - e_f s_{\scriptscriptstyle w}^2 \big)\ , r_{i1}^\sf r_{j1}^\sf + e_f s_{\scriptscriptstyle w}^2 r_{i2}^\sf r_{j2}^\sf \bigg ] \ , . \end{aligned}\ ] ] for the higgs fermion fermion couplings the interaction lagrangian reads @xmath304\ ] ] with the couplings
@xmath305 s_2^t ~=~ - g \frac{m_t \sin\alpha}{2m_{\scriptscriptstyle w}\sin\beta } ~=~ -\frac{h_t}{\sqrt{2}}\sin\alpha , \qquad\qquad & s_2^b ~=~ -g\frac{m_b \cos\alpha}{2m_{\scriptscriptstyle w}\cos\beta } ~=~ -\frac{h_b}{\sqrt{2}}\cos\alpha \ , , \non \\[5 mm ] s_3^t ~=~ ig\frac{m_t\cot\beta}{2m_{\scriptscriptstyle w } } ~=~ i\frac{h_t}{\sqrt{2}}\cos\beta , \qquad\qquad & s_3^b ~=~ ig\frac{m_b\tan\beta}{2m_{\scriptscriptstyle w } } ~=~ i\frac{h_b}{\sqrt{2}}\sin\beta \ , , \non \\[5 mm ] s_4^t ~=~ ig\frac{m_t}{2m_{\scriptscriptstyle w } } ~=~ i\frac{h_t}{\sqrt{2}}\sin\beta , \qquad\qquad & s_4^b ~=~ -ig\frac{m_b}{2m_{\scriptscriptstyle w } } ~=~ -i\frac{h_b}{\sqrt{2}}\cos\beta \ , , \non \\[5 mm ] y_1^t ~=~ g\frac{m_t\cot\beta}{\sqrt2 m_{\scriptscriptstyle w } } ~=~ h_t \cos\beta , \qquad\qquad & y_1^b ~=~ g\frac{m_b\tan\beta}{\sqrt2 m_{\scriptscriptstyle w } } ~=~ h_b \sin\beta \ , , \non \\[5 mm ] y_2^t ~=~ g\frac{m_t}{\sqrt2 m_{\scriptscriptstyle w } } ~=~ h_t \sin\beta , \qquad\qquad & y_2^b ~=~ -g\frac{m_b}{\sqrt2 m_{\scriptscriptstyle w } } ~=~ -h_b \cos\beta \ , .
\non \end{array}\ ] ] the interaction lagrangian for higgs bosons and gauginos is given by @xmath306 \non & & - g \sum_{k=1}^2 h_k^0\ , { \bar{\tilde\chi}}_i^+ \left ( f^+_{ijk } p_r + f^+_{jik } p_l \right ) \chp_j + i g \sum_{k=3}^4 h_k^0\ , { \bar{\tilde\chi}}_i^+ \left ( f^+_{ijk } p_r + f^+_{jik } p_l
\right ) \chp_j \\[2 mm ] & & - g \sum_{k=1}^2 \big [ h_k^+ \ , { \bar{\tilde\chi}}_i^+ \left ( f_{ilk}^r p_r + f_{ilk}^l p_l \right ) \nt_l + { \rm h.c . } \big ] \ , .
\end{aligned}\ ] ] with @xmath307 \\[2 mm ] & & + \frac{d_k}{2 } \big [ z_{l4 } z_{m2 } + z_{m4 } z_{l2 } - \tan\theta_{\scriptscriptstyle w } \left ( z_{l4 } z_{m1 } + z_{m4 } z_{l1 } \right ) \big ] \ = \ f_{mlk}^0 \ , , \\[2 mm ] f_{ijk}^+ & = & \frac{1}{\sqrt2 } \left ( e_k v_{i1 } u_{j2 } - d_k v_{i2 } u_{j1 } \right ) \,,\end{aligned}\ ] ] and @xmath308 \ , , \\[2 mm ] f_{ilk}^l & = & -e_{k+2 } \left [ u_{i1 } z_{l3 } - \frac{1}{\sqrt2 } ( z_{l2 } + z_{l1 } \tan\theta_{\scriptscriptstyle w } ) u_{i2 } \right ] \ , .
\end{aligned}\ ] ] @xmath309 and @xmath310 are rotation matrices which diagonalize the chargino and neutralino mass matrices . @xmath311 and @xmath312 take the values @xmath313 the coupling of the vector boson @xmath314 to two sfermions , @xmath315 with @xmath316 , is given by the matrix @xmath317\non\end{aligned}\ ] ] with @xmath318 and @xmath319 . for the interaction of a vector boson with two gauginos we use the couplings @xmath320\non & o_{ij}^{'l } ~=~ - v_{i1 } v_{j1 } - \frac{1}{2 } v_{i2 } v_{j2 } + \delta_{ij } s_{\scriptscriptstyle
w}^2 \ , , & \\[2mm]\non & o_{ij}^{'r } ~=~ - u_{i1 } u_{j1 } - \frac{1}{2 } u_{i2 } u_{j2 } + \delta_{ij } s_{\scriptscriptstyle w}^2 \ , , & \\[3mm]\non & o_{ij}^{''l } ~=~ - \frac{1}{2 } z_{i3 } z_{j3 } + \frac{1}{2 } z_{i4 } z_{j4 } ~=~ - o_{ij}^{''r } \ , . &
\end{aligned}\ ] ] the interaction lagrangian of the chargino
fermion couplings is given by @xmath321 & & + ~ \overline{\chp_j } \left(l_{ij}^\sb p_l + k_{ij}^\sb p_r \right ) \!t\ , { \sb_i}^{\ast } + \overline{\tilde\chi^{+c}_j } \left(l_{ij}^\st p_l + k_{ij}^\st p_r \right)\ ! b\ , \st_i^\ast\end{aligned}\ ] ] with the coupling matrices @xmath322 k_{ij}^\st & = & h_b u_{j2 } r_{i1}^\st\ , , & k_{ij}^\sb & = & h_t v_{j2 } r_{i1}^\sb\ , .
\end{array}\ ] ] for the neutralino sfermion
fermion couplings the lagrangian reads @xmath323 with the coupling matrices @xmath324 and @xmath325 @xmath326 takes the values @xmath327 for @xmath328down , up@xmath329type case , respectively .
here we give the explicit form of the electroweak contributions to the vertex corrections which are depicted in fig .
[ vertex - graphs ] . for susy qcd contributions we refer to @xcite .
@xmath330 & & + ~{\delta}g_{123}^{\sf ( v , v ) } + { \delta}g_{123}^{\sf ( v , \sf\sf ) } + { \delta}g_{123}^{\sf ( v , h\sf ) } + { \delta}g_{123}^{\sf ( v , az ) } + { \delta}g_{123}^{\sf ( v , ag)}\end{aligned}\ ] ] 1.8 the single contributions correspond to the diagrams with three scalar particles @xmath331 and @xmath332 , three fermions @xmath333 and @xmath334 , one vector particle @xmath335 or two scalar particles @xmath336 and @xmath337 in the loop .
@xmath338 denotes the correction due to the mixing of @xmath0 and @xmath314 and @xmath339 is the higgs mixing transition @xmath0@xmath340 .
as shown in @xcite we can sum up the @xmath341 and @xmath342 transition amplitudes which leads to @xmath343 the explicit form of the @xmath0@xmath314 self energy , @xmath344 , is given in app .
[ appazmixing ] .
the vertex corrections from the exchange of one higgs and two sfermions are @xmath345 \non & & - \frac{1}{(4\pi)^2 } \sum_{m , n = 1}^2 \sum_{k=1}^2 g_{mn3}^{\sf ' } g_{imk}^{\sf\sf ' } g_{jnk}^{\sf\sf'}\ , c_0\big ( m_{\sf_i}^2 , m_{a^0}^2 , m_{\sf_j}^2 , m_{h_k^+}^2 , m_{\sf'_m}^2 , m_{\sf'_n}^2 \big ) \\[-3 mm ] \\[-7 mm ] \non\end{aligned}\ ] ] with the standard two point function @xmath346 @xcite for which we follow the conventions of @xcite .
the graph with 2 higgs particles and one sfermion in the loop leads to @xmath347\non & & \hspace{51 mm } c_0\big ( m_{\sf_i}^2 , m_{a^0}^2 , m_{\sf_j}^2 , m_{\sf_m}^2 , m_{h_k^0}^2 , m_{h_l^0}^2 \big ) \\[2mm]\non & & - \frac{i}{(4\pi)^2 } \ , i_f^{3l } \ , g \ , m_{\scriptscriptstyle w } \sum_{m = 1}^2 \bigg ( g^{\sf\sf'}_{im1 } g^{\sf\sf'}_{jm2}\ , c_0\big ( m_{\sf_i}^2 , m_{a^0}^2 , m_{\sf_j}^2 , m_{\sf'_m}^2 , m_{h^+}^2 , m_{g^+}^2 \big ) \\[2mm]\non & & \hspace{40 mm } -g^{\sf\sf'}_{im2}\ , g^{\sf\sf'}_{jm1}\ , c_0\big ( m_{\sf_i}^2 , m_{a^0}^2 , m_{\sf_j}^2 , m_{\sf'_m}^2 , m_{g^+}^2 , m_{h^+}^2 \big ) \bigg)\end{aligned}\ ] ] with @xmath348 \cos 2\b \ , \cos(\a+\b ) & \sin 2\b \ , \cos(\a+\b ) \end{array } \right)\,.\end{aligned}\ ] ] for the gaugino exchange contributions we get @xmath349\non & & \hspace{-3mm}+\frac{1}{(4\pi)^2 } \sum_{k = 1}^2 f\big ( m_{\sf_i}^2 , m_{a^0}^2 , m_{\sf_j}^2 , m_{\chp_k } , m_{f ' } , m_{f ' } ; s^{f'}_3 , -s^{f'}_3 , k^\sf_{ik } , l^\sf_{ik } , l^\sf_{jk } , k^\sf_{jk } \big)\ , , \\[3mm]\non { \delta}g_{123}^{\sf ( v , f \ti\chi \ti\chi ) } & = & \frac{1}{(4\pi)^2 } \sum_{k , l = 1}^4 \!\ !
f\big ( m_{\sf_i}^2 , m_{a^0}^2 , m_{\sf_j}^2 , m_f , m_{\nt_k } , m_{\nt_l } ; i g f^0_{lk3 } , -i g f^0_{lk3 } , b^\sf_{ik } , a^\sf_{ik } , a^\sf_{jl } , b^\sf_{jl } \big ) \\[1mm]\non & & \hspace{-3mm}+\frac{1}{(4\pi)^2 } \sum_{k , l = 1}^2 \!\ ! f\big ( m_{\sf_i}^2 , m_{a^0}^2 , m_{\sf_j}^2 , m_{f ' } , m_{\chp_k } , m_{\chp_l } ; i g \widetilde f^+_{kl3 } , -i g \widetilde f^+_{lk3 } , k^\sf_{ik } , l^\sf_{ik } , l^\sf_{jl } , k^\sf_{jl } \big)\ , , \\\end{aligned}\ ] ] where @xmath350 shortly stands for @xmath351\non & & \hspace{-9 cm } + ~(h_0 m_0 \!+\ !
m_1)b_0(m_1 ^ 2 , m_0 ^ 2 , m_1 ^ 2 ) + ( h_0m_0 \!+\ ! h_2m_2)b_0(m_2 ^ 2 , m_0 ^ 2 , m_2 ^ 2 ) \\[2mm]\non & & \hspace{-9 cm } + \ , \big[2\!\left(g_0^rg_1^rg_2^r \!+\ ! g_0^lg_1^lg_2^l
\right)\!m_0m_1m_2 + h_0m_0\!\left ( m_1 ^ 2\ ! + \!m_2 ^ 2\!- \!m_0 ^ 2 \right ) + h_1m_1\!\left(m_0 ^ 2\!+\!m_2 ^ 2\!-\!m_2 ^ 2\right ) \\[2 mm ] & & \hspace{-9 cm } + ~h_2m_2\!\left(m_0 ^ 2\!+\!m_1 ^ 2\!-\!m_1 ^ 2\right)\big ] c_0(m_1 ^ 2 , m_0 ^ 2 , m_2 ^ 2 , m_0 ^ 2 , m_1 ^ 2 , m_2 ^ 2)\end{aligned}\ ] ] with the abbreviations @xmath352 and @xmath353 . for
up type sfermions @xmath354 and for down type sfermions
chargino indices are interchanged , @xmath355 .
+ we split the irreducible vertex graphs with one vector particle in the loop into the single contributions of the photon , the @xmath310boson and the @xmath356boson , @xmath357 in order to regularize the infrared divergences we introduce a photon mass @xmath358 .
thus we have @xmath359 where we have used the vector vertex function @xmath360 & & \hspace{-35 mm } + ~\left ( -2m_0 ^ 2 \!+\ !
m_1 ^ 2 \!+\ !
m_2 ^ 2 \!-\ ! m_0 ^ 2 \!+\ !
m_1 ^ 2 \!+\ ! m_2 ^ 2 \right ) c_0\big ( m_1 ^ 2 , m_0 ^ 2 , m_2 ^ 2 , m_0 ^ 2 , m_1 ^ 2 , m_2 ^ 2 \big)\end{aligned}\ ] ] and the rotation matrix @xmath361 , @xmath362 @xmath363 can be found in appendix [ appcouplings ] .
+ for the vertex graphs with 2 sfermions in the loop we obtain @xmath364 b_0 \big ( m_{a^0}^2 , m_{\sf_m}^2 , m_{\sf_n}^2 \big ) \\[1 mm ] \non & & -\frac{1}{(4\pi)^2}\ , g_{\scriptscriptstyle z}^2 \!\ ! \sum_{m , n=1}^2 \!\
! g_{nm3}^\sf \bigg\
{ \bigg [ \big ( \frac{1}{4 } - ( 2 i_f^{3l } \!-\ !
e_f ) e_f s_{\scriptscriptstyle w}^2 \big ) r^{\sf_l}_{ijmn } + e_f^2 s_{\scriptscriptstyle w}^2 r^{\sf_r}_{ijmn } \bigg ] ( n_c^f + 1 ) \\ \non & & \hspace{38 mm } + ( i_f^{3l } \!-\ !
e_f ) e_f s_{\scriptscriptstyle w}^2 \bigg [ n_c^f \big ( r^{\sf}_{ijmn } \!+\ ! r^{\sf}_{mnij } \big ) \!+\ ! r^{\sf}_{inmj } \!+\ !
r^{\sf}_{mjin } \bigg ] \ !
\bigg\ } \\[1 mm ] \non & & \hspace{38 mm } \times \ , b_0 \big ( m_{a^0}^2 , m_{\sf_m}^2 , m_{\sf_n}^2 \big ) \\[1 mm ] \label{vertex2sf } & & -\frac{1}{(4\pi)^2}\ , n_c^{\hat f } \,h_f h_{\hat f}\sum_{m , n=1}^2 \!\ !
g_{nm3}^{\ \hat{\!\!\sf } } \big ( r^{\sf\ \!\hat{\!\!\sf}_{\!f}}_{ijnm } + r^{\sf\ \!\hat{\!\!\sf}_{\!f}}_{jimn } \big ) b_0 \big ( m_{a^0}^2 , m_{\ \!\hat{\!\!\sf}_{\!m}}^2 , m_{\ \!\hat{\!\!\sf}_{\!n}}^2 \big ) \,.\end{aligned}\ ] ] for various products of sfermion rotation matrices we have introduced the short forms @xmath365 r^{\sf_r}_{ijkl } ~=~ r^\sf_{i2 } r^\sf_{j2 } r^\sf_{k2 } r^\sf_{l2 } & r^{\sf\ \!\hat{\!\!\sf}_{\!f}}_{ijkl } ~=~ r^\sf_{i1 } r^\sf_{j2 } r^{\ \!\hat{\!\!\sf}}_{k1 } r^{\ \!\hat{\!\!\sf}}_{l2}\ , .
\end{array}\ ] ] note that the last term in eq .
( [ vertex2sf ] ) originates from the mixing of 2 squarks and 2 sleptons , where @xmath366 denotes the family partner of the fermion @xmath57 with the same isospin and from the same generation , i.e. @xmath367 or @xmath368 .
the diagrams with one higgs boson and one sfermion in the loop lead to @xmath369 \non & & + \frac{i}{(4\pi)^2}\,\sqrt2 \ ,
\sum_{k , m=1}^2 \ !
g^{\sf\sf'}_{imk } \!\left [ \left\ { \!\ ! \begin{array}{ll } ( h_\uparrow^2 \!-\ !
g^2/2 ) \cos^2\b - ( h_\downarrow^2 \!-\ !
g^2/2 ) \sin^2\b \\[2 mm ] ( h_f^2 + h_{f'}^2 - g^2 ) \sin\b \cos\b \end{array } \!\right\}_{\!k } \ ! r^{\sf'}_{m1 } r^\sf_{j1 } \right .
\\[2 mm ] \non & & \hspace{62 mm } + h_f h_{f'}\ , { \delta}_{k2}\ , r^{\sf'}_{m2 } r^\sf_{j2 } \bigg ] b_0 \big ( m_{a^0}^2 , m_{h_k^+}^2 , m_{\sf'_m}^2 \big ) \\[2 mm ] \non & & \ -\
i \leftrightarrow j \,.\end{aligned}\ ] ] with @xmath370 and @xmath371 for the decay into \{squarks , sleptons } , respectively . the higgs
sfermion coupling matrices @xmath372 and @xmath373 can be found in appendix [ appcouplings ] .
( 160,215)(0,0 ) ( 0,-2 )
for the diagonal wave function renormalization constants we use the conventional on shell renormalization conditions which lead to @xmath374 where the dot in @xmath375 denotes the derivative with respect to @xmath376 . in the following we list the single contributions of the wave function corrections ( fig .
[ a0sediag ] and [ sfermionsediag ] ) .
@xmath377 \\[2 mm ] \delta z_{33}^{h,\sf } & = & -\frac{1}{(4\pi)^2}\ , \sum_{f } \sum_{m , n=1}^2 n_c^f\ , g^\sf_{mn3}\ , g^\sf_{nm3}\ , \dot{b}_0 ( m_{a^0}^2 , m_{\sf_m}^2 , m_{\sf_n}^2 ) \\[2mm]\non \delta z_{33}^{h,\nt } & = & \frac{1}{(4\pi)^2}\ , g^2 \sum_{k , l=1}^4 ( f^0_{kl3})^2 \bigg [ \dot{b}_0(m_{a^0}^2 , m_{\nt_k}^2 , m_{\nt_l}^2 ) \big ( ( m_{\nt_k}-m_{\nt_l})^2 - m_{a^0}^2 \big ) \\ & & \hspace{60 mm } - b_0(m_{a^0}^2 , m_{\nt_k}^2 , m_{\nt_l}^2 ) \bigg ] \\[2mm]\non \delta z_{33}^{h,\chp } & = & \frac{1}{(4\pi)^2}\ , g^2 \sum_{k , l=1}^2 \bigg [ \big ( ( f^+_{kl3})^2 + ( f^+_{lk3})^2 \big ) \bigg ( \ ! \big ( m_{\chp_k}^2 \!+\ ! m_{\chp_l}^2 \!-\ ! m_{a^0}^2
\big ) \dot{b}_0 - b_0 \bigg ) \\ & & \hspace{40 mm } - 4 m_{\chp_k } m_{\chp_l } f^+_{kl3 } \ , f^+_{lk3 } \ , \dot{b}_0 \bigg ] ( m_{a^0}^2 , m_{\chp_k}^2 , m_{\chp_l}^2 ) \\[2mm]\non \delta z_{33}^{h , h } & = & -\frac{1}{(4\pi)^2}\ , \big(\frac{g_{\scriptscriptstyle z}\,m_{\scriptscriptstyle z}}{2}\big)^2 \ , \sum_{k=1}^2 \sum_{l=3}^4 \big ( a_{k , l-2 } \big)^2\ , \dot{b}_0(m_{a^0}^2 , m_{h_k^0}^2 , m_{h_l^0}^2 ) \\[2 mm ] & & \hspace{40mm}- \frac{1}{(4\pi)^2}\ , 2\ , \big(\frac{g\,m_{\scriptscriptstyle w}}{2}\big)^2\ , \dot{b}_0(m_{a^0}^2 , m_{h^+}^2 , m_{w^+}^2 ) \\[2mm]\non \delta z_{33}^{h , z } & = & \frac{1}{(4\pi)^2}\ , \frac{g_{\scriptscriptstyle z}^2}{4 } \sum_{k=1}^2 \big ( r_{1k}(\a\!-\!\b ) \big)^2 \bigg [ \dot{b}_0(m_{a^0}^2 , m_{h_k^0}^2 , m_{z}^2 ) \big ( 2 m_{a^0}^2 \!+\ ! 2 m_{h_k^0}^2 \!-\ ! m_{\scriptscriptstyle z}^2 \big ) \\ & & \hspace{60mm}+2 b_0(m_{a^0}^2 , m_{h_k^0}^2 , m_{z}^2 ) \bigg ] \\[2mm]\non \delta z_{33}^{h , w } & = & \frac{1}{(4\pi)^2}\,2\ , \frac{g^2}{4 } \bigg [ \dot{b}_0(m_{a^0}^2 , m_{h^+}^2 , m_{w}^2 ) \big ( 2 m_{a^0}^2 \!+\ ! 2 m_{h^+}^2 \!-\ ! m_{\scriptscriptstyle w}^2 \big ) \\ & & \hspace{60mm}+2 b_0(m_{a^0}^2 , m_{h^+}^2 , m_{w}^2 ) \bigg]\end{aligned}\ ] ] @xmath378\non & & \hspace{45 mm } + 4 m_{\nt_k } m_f \ , a^\sf_{ik } b^\sf_{ik } \ , \dot{b}_0 \bigg ] ( m_{\sf_i}^2 , m_{\nt_k}^2 , m_f^2 ) \\[2mm]\non & & \hspace{-3mm}+\vor \ , \sum_{k=1}^2 \bigg [ \big ( ( k^\sf_{ik})^2 + ( l^\sf_{ik})^2 \big ) \cdot \big ( ( m_{\chp_k}^2 + m_{f'}^2 - m_{\sf_i}^2)\ , \dot{b}_0 - b_0 \big ) \\[2 mm ] & & \hspace{45 mm } + 4 m_{\chp_k } m_{f ' } \ , k^\sf_{ik } l^\sf_{ik } \ , \dot{b}_0 \bigg ] ( m_{\sf_i}^2 , m_{\chp_k}^2 , m_{f'}^2 ) \\[2mm]\non \delta z_{ii}^{\sf,\ , h } & = & -\vor \ , \sum_{k=1}^4 \sum_{m=1}^2 g^\sf_{mik}\ , g^\sf_{imk}\ , \dot{b}_0(m_{\sf_i}^2 , m_{\sf_m}^2 , m_{h_k^0}^2 ) \\[2 mm ] & & -\vor \ , \sum_{k=1}^2 \sum_{m=1}^2 g^{\sf'\sf}_{mik } \ , g^{\sf\sf'}_{imk}\ , \dot{b}_0(m_{\sf_i}^2 , m_{\sf'_m}^2 , m_{h_k^+}^2 ) \\[2mm]\non \delta z_{ii}^{\sf,\ , \g } & = & \vor \ , ( e_0 e_f)^2 \bigg [ 2 b_0 + \big ( 4 m_{\sf_i}^2 \!-\ ! \l^2
\big ) \dot{b}_0 \bigg ] ( m_{\sf_i}^2 , m_{\sf_i}^2 , \l^2 ) \\[2 mm ] \delta z_{ii}^{\sf,\ , z } & = & \vor \ , g_{\scriptscriptstyle
z}^2 \sum_{m=1}^2 \big(z^\sf_{im}\big)^2 \bigg [ 2 b_0 + \big ( 2 m_{\sf_i}^2 \!+\ ! 2 m_{\sf_m}^2 \!-\ !
m_{\scriptscriptstyle z}^2
\big ) \dot{b}_0 \bigg ] ( m_{\sf_i}^2 , m_{\sf_m}^2 , m_{\scriptscriptstyle z}^2 ) \\[2mm]\non \delta z_{ii}^{\sf,\ , w } & = & \vor \ , \frac{g^2}{2 } \sum_{m=1}^2 \big ( r^{\sf}_{i1 } r^{\sf'}_{m1 } \big)^2 \bigg [ 2 b_0 + \big ( 2 m_{\sf_i}^2 \!+\ ! 2 m_{\sf'_m}^2 \!-\ !
\big ) \dot{b}_0 \bigg ] ( m_{\sf_i}^2 , m_{\sf'_m}^2 , m_{\scriptscriptstyle w}^2 ) \\\end{aligned}\ ] ] ( 170,105)(0,0 ) ( 0,0 ) ( 170,70)(0,0 ) ( 0,0 )
here we give the explicit form of the self
energies needed for the computation of various counter terms for the one loop width @xmath379 . + the single contributions from the particles @xmath326 are denoted by the superscript @xmath326 in @xmath381 .
@xmath382 \pi_{az}^{\nt } & = & \frac{i}{(4\pi)^2}\ , 2\ , g\ , g_{\scriptscriptstyle z } \sum_{k , l=1}^4 \!f^0_{kl3}\ , o^{''l}_{lk } \big [ m_{\nt_l}\ , b_0 + ( m_{\nt_l}\!-\!m_{\nt_k } ) b_1 \big ] ( m_{a^0}^2 , m_{\nt_l}^2 , m_{\nt_k}^2 ) \\[2mm]\non \pi_{az}^{\chp } & = & \frac{i}{(4\pi)^2}\ , 2\ , g\ , g_{\scriptscriptstyle z } \sum_{k , l=1}^2 \bigg [ \big ( f^+_{kl3}\ , o^{'l}_{lk } - f^+_{lk3}\ , o^{'r}_{lk } \big ) m_{\chp_l } ( b_0 + b_1 ) \\ & & \hspace{35mm}+ \big ( f^+_{kl3}\ , o^{'r}_{lk } - f^+_{lk3}\ , o^{'l}_{lk } \big ) m_{\chp_k}\ , b_1 \bigg ] ( m_{a^0}^2 , m_{\chp_l}^2 , m_{\chp_k}^2 ) \\[2 mm ] \pi_{az}^\sf & = & -\vor\ , 2\,g_{\scriptscriptstyle z } \sum_{f } n_c^f\,z^\sf_{21}\ , g^\sf_{123 } \big ( b_0 + 2 b_1 \big ) ( m_{a^0}^2 , m_{\sf_1}^2 , m_{\sf_2}^2 ) \\[2 mm ] \pi_{az}^h & = & \frac{i}{(4\pi)^2}\ , \frac{g_{\scriptscriptstyle z}^2\ , m_{\scriptscriptstyle z}}{4 } \sum_{k=1}^2 \sum_{l=3}^4 a_{k , l-2}\ , r_{k , l-2}(\b\!-\!\a ) \big ( b_0 + 2 b_1 \big ) ( m_{a^0}^2 , m_{h_l^0}^2 , m_{h_k^0}^2 ) \\[2 mm ] \pi_{az}^z & = & \frac{i}{(4\pi)^2}\ , \frac{g_{\scriptscriptstyle z}^2\ , m_{\scriptscriptstyle z}}{4 } \sin(2\a\!-\!2\b ) \sum_{k=1}^2 ( -1)^k \big ( b_0 - b_1 \big ) ( m_{a^0}^2 , m_{h_k^0}^2 , m_{\scriptscriptstyle z}^2 ) \end{aligned}\ ] ] for the calculation of the mass counter term of a gauge boson @xmath383 @xmath384 , @xmath385 , we need the transverse part of the vector self energy @xmath386 from @xmath387 @xmath388 \\[2 mm ] \left ( \frac{{\delta}m_{\scriptscriptstyle w}}{m_{\scriptscriptstyle w } } \right)^{\sf\sf } & = & -\frac{1}{(4 \pi)^2 } \ , \frac{g^2}{m_{\scriptscriptstyle w}^2 } \sum_{\rm gen .
} n_c^f \sum_{m , n = 1}^2 \big ( r^{\sf_{\uparrow}}_{m1 } r^{\sf_{\downarrow}}_{n1 } \big)^2 b_{00}(m_{\scriptscriptstyle w}^2 , m_{\sf_{\uparrow m}}^2 , m_{\sf_{\downarrow n}}^2 ) \end{aligned}\ ] ] here @xmath389 and @xmath390 denote up and down type ( s)fermions of all three generations , respectively .
@xmath391 \non \left ( \frac{{\delta}m_{\scriptscriptstyle w}}{m_{\scriptscriptstyle w } } \right)^{\ti\chi } & = & \vor \ , \frac{g^2}{m_{\scriptscriptstyle w}^2 } \sum_{k=1}^2 \sum_{l=1}^4 \bigg [ 2\ , o^l_{lk}\ , o^r_{lk}\ , m_{\chp_k}\ , m_{\nt_l}\ , b_0 - \big ( ( o^l_{lk})^2 + ( o^r_{lk})^2 \big)\ , \times \\ & & \hspace{7 mm } \big ( m_{\scriptscriptstyle w}^2\ , b_1 + m_{\nt_l}^2\ , b_0 + a_0(m_{\chp_k}^2 ) - 2 \ , b_{00 } \big ) \bigg ] ( m_{\scriptscriptstyle w}^2 , m_{\nt_l}^2 , m_{\chp_k}^2 ) \\[2mm]\non \left ( \frac{{\delta}m_{\scriptscriptstyle w}}{m_{\scriptscriptstyle w } } \right)^{hh } & = & -\vor \ , \frac{g^2}{2 m_{\scriptscriptstyle w}^2 } \bigg[\ , \sum_{k , l=1}^2 \big ( r_{lk}(\a\!-\!\b ) \big)^2\ , b_{00}(m_{\scriptscriptstyle w}^2 , m_{h_l^+}^2 , m_{h_k^0}^2 ) \\ & & \hspace{25 mm } + ~b_{00}(m_{\scriptscriptstyle w}^2 , m_{h^+}^2 , m_{a^0}^2 ) + b_{00}(m_{\scriptscriptstyle w}^2 , m_{g^+}^2 , m_{g^0}^2 ) \bigg ] \\[2 mm ] \left ( \frac{{\delta}m_{\scriptscriptstyle w}}{m_{\scriptscriptstyle w } } \right)^{h } & = & \vor \ , \frac{g^2}{8 m_{\scriptscriptstyle w}^2 } \left(\ \sum_{k=1}^4 a_0(m_{h_k^0}^2 ) + 2 \sum_{k=1}^2 a_0(m_{h_k^+}^2 ) \right ) \\[2mm]\non \left ( \frac{{\delta}m_{\scriptscriptstyle w}}{m_{\scriptscriptstyle w } } \right)^{vs } & = & \vor \ , \frac{g^2}{2 } \bigg [ \ \sum_{k=1}^2 \big ( r_{2k}(\a\!-\!\b ) \big)^2 b_0(m_{\scriptscriptstyle w}^2 , m_{h_k^0}^2 , m_{\scriptscriptstyle w}^2 ) \\ & & \hspace{15 mm } + ~s_{\scriptscriptstyle w}^2 \ , b_0(m_{\scriptscriptstyle w}^2 , m_{\scriptscriptstyle w}^2 , \l^2 ) + s_{\scriptscriptstyle w}^2 t_{\scriptscriptstyle w}^2 \ , b_0(m_{\scriptscriptstyle w}^2 , m_{\scriptscriptstyle w}^2 , m_{\scriptscriptstyle z}^2 ) \bigg ] \\[2mm]\non \left ( \frac{{\delta}m_{\scriptscriptstyle w}}{m_{\scriptscriptstyle w } } \right)^{vv+v+\rm ghost } \hspace{-2mm}&= & -\vor \ , \frac{g^2}{2 m_{\scriptscriptstyle w}^2 } \bigg [ s_{\scriptscriptstyle w}^2 \bigg(\ ! 8 b_{00 }
+ 7 m_{\scriptscriptstyle w}^2 b_0 + 2 m_{\scriptscriptstyle w}^2 b_1 \ !
\bigg ) ( m_{\scriptscriptstyle w}^2 , m_{\scriptscriptstyle w}^2 , \l^2 ) \\ \non & & \hspace{23 mm } + c_{\scriptscriptstyle w}^2 \bigg(\ ! 8 b_{00 } + 7 m_{\scriptscriptstyle w}^2 b_0 + 2 m_{\scriptscriptstyle w}^2 b_1 \ !
\bigg ) ( m_{\scriptscriptstyle w}^2 , m_{\scriptscriptstyle w}^2 , m_{\scriptscriptstyle z}^2 ) \\ & & \hspace{35 mm } - s_{\scriptscriptstyle w}^2 a_0(\l^2 ) - c_{\scriptscriptstyle w}^2 a_0(m_{\scriptscriptstyle z}^2 ) - 3 a_0(m_{\scriptscriptstyle w}^2 ) \bigg]\end{aligned}\ ] ] accordingly to eq .
( [ vvse_def ] ) the mass counter term contributions to the gauge boson @xmath314 are : @xmath392 \big ( m_{\scriptscriptstyle z}^2 , m_f^2 , m_f^2 \big ) \\[2 mm ] \left ( \frac{{\delta}m_{\scriptscriptstyle z}}{m_{\scriptscriptstyle z } } \right)^{\sf\sf } & = & -\vor \ , \frac{2\ , g_{\scriptscriptstyle z}^2}{m_{\scriptscriptstyle z}^2 } \sum_{f } n_c^f \sum_{m , n = 1}^2 \big ( z^\sf_{mn } \big)^2 \ , b_{00}(m_{\scriptscriptstyle z}^2 , m_{\sf_m}^2 ,
m_{\sf_n}^2 ) \\[2 mm ] \left ( \frac{{\delta}m_{\scriptscriptstyle z}}{m_{\scriptscriptstyle z } } \right)^{\sf } & = & \vor \ , \frac{g_{\scriptscriptstyle z}^2}{m_{\scriptscriptstyle z}^2 } \sum_{f } n_c^f \sum_{m=1}^2 \bigg ( \big(c^f_l\,r^\sf_{m1}\big)^2 + \big(c^f_r \,r^\sf_{m2}\big)^2 \bigg ) a_0(m_{\sf_{m}}^2 ) \\[2mm]\non \left ( \frac{{\delta}m_{\scriptscriptstyle z}}{m_{\scriptscriptstyle z } } \right)^{\nt } & = & -\vor \ ,
\frac{g_{\scriptscriptstyle z}^2}{m_{\scriptscriptstyle z}^2 } \sum_{k , l=1}^4 \big ( o^{''}_{kl } \big)^2 \bigg [ ( m_{\nt_k } \!+\ !
m_{\nt_l } ) m_{\nt_l}\ , b_0 + m_{\scriptscriptstyle z}^2\ , b_1 \\ & & \hspace{40 mm } + a_0(m_{\nt_k}^2 ) -~2\ , b_{00 } \bigg ] ( m_{\scriptscriptstyle z}^2 , m_{\nt_k}^2 , m_{\nt_l}^2 ) \\[2mm]\non \left ( \frac{{\delta}m_{\scriptscriptstyle z}}{m_{\scriptscriptstyle z } } \right)^{\chp } & = & \vor \ , \frac{g_{\scriptscriptstyle z}^2}{m_{\scriptscriptstyle z}^2 } \sum_{k , l=1}^2
\bigg [ 2\ , o^{'l}_{kl}\ , o^{'r}_{kl}\ , m_{\chp_k}\ , m_{\chp_l}\ , b_0 - \big ( ( o^{'l}_{kl})^2 + ( o^{'r}_{kl})^2 \big)\ , \times \\ & & \hspace{4 mm } \big ( m_{\scriptscriptstyle z}^2\ , b_1 + m_{\chp_k}^2\ , b_0 + a_0(m_{\chp_k}^2 ) - 2 \ , b_{00 } \big ) \bigg ] ( m_{\scriptscriptstyle z}^2 , m_{\chp_k}^2 , m_{\chp_l}^2 ) \\[2mm]\non \left ( \frac{{\delta}m_{\scriptscriptstyle z}}{m_{\scriptscriptstyle z } } \right)^{hh } & = & -\vor \ , \frac{g_{\scriptscriptstyle z}^2}{2 m_{\scriptscriptstyle z}^2 } \bigg [ \sum_{k=1}^2 \sum_{l=3}^4 \big ( r_{k , l-2}(\b\!-\!\a ) \big)^2 \ , b_{00}(m_{\scriptscriptstyle z}^2 , m_{h_k^0}^2 , m_{h_l^0}^2 ) \\ & & \hspace{37 mm } + \cos^2(2\tw ) \sum_{k=1}^2 b_{00}(m_{\scriptscriptstyle z}^2 , m_{h_k^+}^2 , m_{h_k^+}^2 ) \bigg ] \\[2 mm ] \left ( \frac{{\delta}m_{\scriptscriptstyle z}}{m_{\scriptscriptstyle z } } \right)^{h } & = & \vor \ , \frac{g_{\scriptscriptstyle z}^2}{8 m_{\scriptscriptstyle z}^2 } \bigg [ \sum_{k=1}^4 a_0(m_{h_k^0}^2 ) + 2 \cos^2(2 \tw ) \sum_{k=1}^2 a_0(m_{h_k^+}^2 ) \bigg ] \\[2mm]\non \left ( \frac{{\delta}m_{\scriptscriptstyle z}}{m_{\scriptscriptstyle z } } \right)^{vs } & = & \vor \bigg ( \frac{g_{\scriptscriptstyle z}^2}{2 } \sin^2(\a\!-\!\b)\ , b_0(m_{\scriptscriptstyle z}^2 , m_{h^0}^2 , m_{\scriptscriptstyle z}^2 ) + \frac{g_{\scriptscriptstyle z}^2}{2 } \cos^2(\a\!-\!\b ) \ , \times \\ & & \hspace{21 mm } b_0(m_{\scriptscriptstyle z}^2 , m_{h^0}^2 , m_{\scriptscriptstyle z}^2 ) + g^2\ , s_{\scriptscriptstyle w}^4 \ , b_0(m_{\scriptscriptstyle z}^2 , m_{w}^2 , m_{g^+}^2 ) \bigg ) \\[2mm]\non \left ( \frac{{\delta}m_{\scriptscriptstyle z}}{m_{\scriptscriptstyle z } } \right)^{ww+w+ \rm ghost } & = & -\vor \ , \frac{g^2 c_{\scriptscriptstyle w}^2}{m_{\scriptscriptstyle z}^2 } \bigg [ 4 b_{00 } + m_{\scriptscriptstyle w}^2 b_0 + \frac{5}{2 } m_{\scriptscriptstyle z}^2 b_0 + m_{\scriptscriptstyle z}^2 b_1 \\ & & \hspace{40 mm } - 2 a_0(m_{\scriptscriptstyle w}^2 ) \ !
\bigg](m_{\scriptscriptstyle z}^2 , m_{\scriptscriptstyle w}^2 , m_{\scriptscriptstyle w}^2)\end{aligned}\ ] ] + with @xmath393 thus the counter term for quarks and leptons is given by @xmath394 \ , .
\end{aligned}\ ] ] note that for quarks and leptons ( contrary to charginos ) , the left and right handed scalar parts of @xmath395 are equal , @xmath396 .
the single contributions to @xmath397 are as follows : @xmath398 ( m_f^2 , m_f^2 , m_{h_k^0}^2 ) \\[-3 mm ] \\
\non \bigg ( \frac{{\delta}m_f}{m_f } \bigg)^{f ' h_k^+ } & = & -\vor \sum_{k=1}^2 \bigg [ \frac{1}{2 } \bigg ( ( y^f_{k})^2 + ( y^{f'}_{k})^2 \bigg ) b_1 - \frac{m_{f'}}{m_f}\ , y^f_{k}\ , y^{f'}_{k } \,b_0 \bigg ] ( m_f^2 , m_{f'}^2 , m_{h_k^+}^2 ) \\ \\[2mm]\non \bigg ( \frac{{\delta}m_f}{m_f } \bigg)^{\sf\,\nt } & = & -\vor \sum_{m=1}^2 \sum_{k=1}^4 \bigg [ \frac{1}{2 } \bigg ( ( a^\sf_{mk})^2 + ( b^\sf_{mk})^2 \bigg ) b_1 \\ & & \hspace{40 mm } - \frac{m_{\nt_k}}{m_f}\ , a^\sf_{mk } b^\sf_{mk } \,b_0 \bigg ] ( m_f^2 , m_{\nt_k}^2 , m_{\sf_m}^2 ) \\[2mm]\non \bigg ( \frac{{\delta}m_f}{m_f } \bigg)^{\sf'\chp } & = & -\vor \sum_{m=1}^2 \sum_{k=1}^2 \bigg [ \frac{1}{2 } \bigg ( ( k^{\sf'}_{mk})^2 + ( l^{\sf'}_{mk})^2 \bigg ) b_1 \\ & & \hspace{40 mm } - \frac{m_{\chp_k}}{m_f}\ , k^{\sf'}_{mk}\ , l^{\sf'}_{mk } \,b_0 \bigg ] ( m_f^2 , m_{\chp_k}^2 , m_{\sf'_m}^2 ) \\[2mm]\non \bigg ( \frac{{\delta}m_f}{m_f } \bigg)^{f \g } & = & -\vor \ , 2 ( e_0 e_f)^2 \big ( b_0 - b_1 \big ) ( m_f^2 , \l^2 , m_{f}^2 ) \\[2 mm ] \bigg ( \frac{{\delta}m_f}{m_f } \bigg)^{f z^0 } & = & -\vor \ , g_{\scriptscriptstyle z}^2 \bigg [ \big ( ( c^f_l)^2 + ( c^f_r)^2 \big ) b_1 + 4 c^f_l c^f_r \ , b_0 \bigg ] ( m_f^2 , m_f^2 , m_{\scriptscriptstyle z}^2 ) \\[2 mm ] \bigg ( \frac{{\delta}m_f}{m_f } \bigg)^{f ' w^+ } & = & -\vor \ , \frac{g^2}{2 } \ , b_1 ( m_f^2 , m_{f'}^2 , m_{\scriptscriptstyle w}^2 ) \end{aligned}\ ] ] since the higgsino mass parameter @xmath54 is fixed in the chargino sector , the counter term @xmath399 reads @xcite @xmath400 with the chargino self
energies @xmath401 .
@xmath402 and @xmath383 are two real @xmath403 matrices which diagonalize the chargino mass matrix , @xmath404 the single left and right handed parts of @xmath405 can be found by comparing the coefficients accordingly to eq .
( [ piferm ] ) .
+ fermion
sfermion contribution : @xmath406 \\[2 mm ] \non \pi_{ij}^{r , f}(k^2 ) & = & -\vor \ , \sum_f n_c^f \sum_{m=1}^2 \bigg [ l_{mi}^{\sf_\downarrow } l_{mj}^{\sf_\downarrow}\ , b_1\!\big(k^2 , m_{f_\uparrow}^2 , m_{\sf_{\downarrow m}}^2\big ) + k_{mi}^{\sf_\uparrow } k_{mj}^{\sf_\uparrow } \,b_1\!\big(k^2 , m_{f_\downarrow}^2 , m_{\sf_{\uparrow m}}^2\big ) \bigg ] \\[2 mm ] \non \pi_{ij}^{sl , f}(k^2 ) & = & \hspace{3.25 mm } \vor \ , \sum_f n_c^f
\sum_{m=1}^2 \bigg [ m_{f_{\uparrow } } l_{mi}^{\sf_\downarrow } k_{mj}^{\sf_\downarrow}\ , b_0\big(k^2 , m_{f_\uparrow}^2 , m_{\sf_{\downarrow m}}^2\big ) + m_{f_{\downarrow } } k_{mi}^{\sf_\uparrow } l_{mj}^{\sf_\uparrow } \ , \times \\[-2 mm ] \non & & \hspace{96.5 mm } b_0\big(k^2 , m_{f_\downarrow}^2 , m_{\sf_{\uparrow m}}^2\big ) \bigg ] \\[2 mm ] \non \pi_{ij}^{sr , f}(k^2 ) & = & \hspace{3.25 mm } \vor \ , \sum_f n_c^f \sum_{m=1}^2 \bigg [ m_{f_{\uparrow } } k_{mi}^{\sf_\downarrow } l_{mj}^{\sf_\downarrow}\ , b_0\big(k^2 , m_{f_\uparrow}^2 , m_{\sf_{\downarrow m}}^2\big ) + m_{f_{\downarrow } } l_{mi}^{\sf_\uparrow } k_{mj}^{\sf_\uparrow } \ , \times \\[-2 mm ] \non & & \hspace{96.5 mm } b_0\big(k^2 , m_{f_\downarrow}^2 , m_{\sf_{\uparrow m}}^2\big ) \bigg ] \\[2mm]\end{aligned}\ ] ] higgs / gaugino contribution : @xmath407 \ ! \big(k^2 , m_{\chp_k}^2 , m_{h_l^0}^2\big ) \\\end{aligned}\ ] ] @xmath408(k^2 , m_{\nt_k}^2 , m_{h_l^+}^2\big ) \\\end{aligned}\ ] ] @xmath409(k^2 , m_{\chp_j}^2 , \l^2)\end{aligned}\ ] ] @xmath410(k^2 , m_{\chp_k}^2 , m_{\scriptscriptstyle z}^2\big ) \\\end{aligned}\ ] ] @xmath411(k^2 , m_{\nt_k}^2 , m_{\scriptscriptstyle w}^2\big ) \\\end{aligned}\ ] ] for the fixing of the sfermion mixing angle @xmath412 we need the off diagonal elements of the sfermion self energies , @xmath413 . in the following , @xmath414 denotes the weak hypercharge , @xmath415 .
the short forms for various products of sfermion rotation matrices can be found in [ appvertex ] .
additionally , we use the abbreviation @xmath416 .
@xmath417\non & & \hspace{20 mm } + \big ( a^\sf_{ik } b^\sf_{jk } + b^\sf_{ik } a^\sf_{jk } \big ) \cdot 2 m_{\nt_k } m_f b_0 \bigg ] ( m_{\sf_j}^2 , m_{\nt_k}^2 , m_f^2 ) \\[2mm]\non & & -\vor \ , \sum_{k=1}^2 \bigg [ \big ( k^\sf_{ik } k^\sf_{jk } + l^\sf_{ik } l^\sf_{jk } \big ) \cdot \big ( a_0(m_{\chp_k}^2 ) + a_0(m_{f'}^2 ) + ( m_{\chp_k}^2 + m_{f'}^2 - m_{\sf_j}^2 ) \ , b_0 \big ) \\[2 mm ] & & \hspace{20 mm } + \big ( k^\sf_{ik } l^\sf_{jk } + l^\sf_{ik } k^\sf_{jk } \big ) \cdot 2 m_{\chp_k } m_{f ' } b_0 \bigg ] ( m_{\sf_j}^2 , m_{\chp_k}^2 , m_{f'}^2)\end{aligned}\ ] ] @xmath418 & & \hspace{-3mm}+\vor \ , \sum_{k=1}^2 \sum_{m=1}^2 g^{\sf'\sf}_{mik } \ , g^{\sf\sf'}_{jmk}\ , b_0(m_{\sf_j}^2 , m_{\sf'_m}^2 , m_{h_k^+}^2 ) \\[2mm]\non \pi_{ij}^{\sf,\ , \g\sf } & = & -\vor \ , ( e_0 e_f)^2 { \delta}_{ij } \bigg [ 2 a_0(\l^2 ) - a_0(m_{\sf_i}^2 ) + \big ( 4 m_{\sf_i}^2 \!-\ !
\l^2 \big ) b_0 ( m_{\sf_i}^2 , m_{\sf_i}^2 , \l^2 ) \bigg ] \\
\\[2mm]\non \pi_{ij}^{\sf,\ , z\sf } & = & -\vor \ , g_{\scriptscriptstyle z}^2 \sum_{m=1}^2 z^\sf_{mi } z^\sf_{jm } \bigg [ 2 a_0(m_{\scriptscriptstyle
z}^2 ) -a_0(m_{\sf_m}^2 ) + \big ( 2 m_{\sf_j}^2 \!+\ ! 2 m_{\sf_m}^2 \!-\ !
m_{\scriptscriptstyle z}^2
\big ) \ , \times \\ & & \hspace{87mm}b_0 ( m_{\sf_j}^2 , m_{\sf_m}^2 , m_{\scriptscriptstyle z}^2 ) \bigg ] \\[2mm]\non \pi_{ij}^{\sf,\ , w\sf ' } & = & -\vor \ , \frac{g^2}{2 } r^{\sf}_{i1 } r^{\sf}_{j1 } \!\sum_{m=1}^2\
! \big ( r^{\sf'}_{m1 } \big)^2 \bigg [ 2 a_0(m_{\scriptscriptstyle w}^2 ) -a_0(m_{\sf'_m}^2 ) + \big ( 2 m_{\sf_j}^2 \!+\ !
2 m_{\sf'_m}^2 \!-\ !
\big ) \ , \times \\ & & \hspace{87mm}b_0 ( m_{\sf_j}^2 , m_{\sf'_m}^2 , m_{\scriptscriptstyle w}^2 ) \bigg]\end{aligned}\ ] ] @xmath419 a_0(m_{\sf_m}^2 ) \\[1 mm ] \non & & \hspace{-3mm}+\frac{1}{(4\pi)^2}\ , g_{\scriptscriptstyle z}^2 \sum_{m=1}^2 \bigg\ { \bigg [ \big ( \frac{1}{4 } - ( 2 i_f^{3l } \!-\ !
e_f ) e_f s_{\scriptscriptstyle w}^2 \big ) r^{\sf_l}_{jimm } + e_f^2 s_{\scriptscriptstyle w}^2 r^{\sf_r}_{jimm } \bigg ] ( n_c^f + 1 ) \\ \non & & \hspace{25 mm } + ( i_f^{3l } \!-\ !
e_f ) e_f s_{\scriptscriptstyle w}^2 \bigg [ n_c^f \big ( r^{\sf}_{jimm } \!+\ ! r^{\sf}_{mmji } \big ) \!+\ ! r^{\sf}_{jmmi } \!+\ !
r^{\sf}_{mijm } \bigg ] \ !
\bigg\ } a_0(m_{\sf_m}^2 ) \\
\\\non \pi_{ij}^{\sf,\ , \sf ' } & = & \frac{1}{(4\pi)^2 } \sum_{m=1}^2 \big ( h_f^2 \ , r^{\sf'\!\sf_d}_{mmji } + h_{f'}^2\ , r^{\sf\sf'_d}_{jimm } \big ) a_0(m_{\sf'_m}^2 ) \\[1 mm ] \non & & \hspace{-3mm}+\frac{1}{(4\pi)^2}\ , \frac{g^2}{4 } \sum_{m=1}^2 \bigg\ { n_c^f \bigg [ \big ( t_{\scriptscriptstyle w}^2 y^f_l y^{f'}_l \!-\ ! 1 \big ) r^{\sf\sf'_l}_{jimm } + t_{\scriptscriptstyle w}^2 y^f_r y^{f'}_r \ , r^{\sf\sf'_r}_{jimm } \\[1 mm ] & & \hspace{28mm}- y^f_l y^{f'}_r r^{\sf\sf'_d}_{jimm } - y^{f'}_l y^f_r r^{\sf'\!\sf_d}_{mmji } \bigg ] + 2 r^{\sf\sf'_l}_{jimm } \bigg\ } a_0(m_{\sf'_m}^2 ) \\[2mm]\non \pi_{ij}^{\sf,\ , \
\!\hat{\!\!\sf } } & = & \frac{1}{(4\pi)^2 } n_c^{\hat f } \sum_{m=1}^2 \bigg [ h_f h_{\hat f } \big
( r^{\sf\ \!\hat{\!\!\sf}_{\!f}}_{ijmm } + r^{\sf\ \!\hat{\!\!\sf}_{\!f}}_{jimm } \big ) + \frac{g^2}{4 } r^{\sf\ \!\hat{\!\!\sf}_{\!l}}_{jimm } \bigg ] a_0(m_{\ \!\hat{\!\!\sf}_{\!m}}^2 ) \\[1 mm ] \non & & \hspace{-3mm}+\frac{1}{(4\pi)^2}\ , n_c^{\hat f}\ , \frac{g'^2}{4 } \sum_{m=1}^2 \bigg [ y^f_l y^{\hat{f}}_l r^{\sf\ \!\hat{\!\!\sf}_{\!l}}_{jimm } - y^f_l y^{\hat{f}}_r \,r^{\sf\ \!\hat{\!\!\sf}_{\!d}}_{jimm } - y^{\hat{f}}_l y^f_r \,r^{\ \hat{\!\!\sf}\sf_d}_{mmji } \\[1 mm ] & & \hspace{70 mm } + ~y^f_r y^{\hat{f}}_r r^{\sf\ \!\hat{\!\!\sf}_{\!r}}_{jimm } \bigg ] a_0(m_{\ \!\hat{\!\!\sf}_{\!m}}^2 ) \\[2mm]\non \pi_{ij}^{\sf,\ , \ \!\hat{\!\!\sf}\,\ ! ' } & = & -\frac{1}{(4\pi)^2 } n_c^{\hat f}\ , \frac{g^2}{4 } \sum_{m=1}^2 r^{\sf\ \!\hat{\!\!\sf}\,\!'_{\!l}}_{jimm } a_0(m_{\
\!\hat{\!\!\sf}\,\!'_{\!m}}^2 ) \\[1 mm ] \non & & + \frac{1}{(4\pi)^2}\ , n_c^{\hat f}\ , \frac{g'^2}{4 } \sum_{m=1}^2 \bigg [ y^f_l y^{\hat{f}'}_l r^{\sf\ \!\hat{\!\!\sf}\,\!'_{\!l}}_{jimm } - y^f_l y^{\hat{f}'}_r \,r^{\sf\ \!\hat{\!\!\sf}\,\!'_{\!d}}_{jimm } - y^{\hat{f}'}_l y^f_r \,r^{\ \hat{\!\!\sf}\,\!'\!\sf_d}_{mmji } \\[1 mm ] & & \hspace{70 mm } + ~y^f_r y^{\hat{f}'}_r r^{\sf\ \!\hat{\!\!\sf}\,\!'_{\!r}}_{jimm } \bigg ] a_0(m_{\ \!\hat{\!\!\sf}\,\!'_{\!m}}^2)\end{aligned}\ ] ] the contributions from first and second generation sfermions , @xmath420 , are given by @xmath421 \pi_{ij}^{\sf,\ , \ti f ' } & = & \pi_{ij}^{\sf,\ , \ \!\hat{\!\!\sf}\,\ ! ' } ( \hat f ' \rightarrow f')\ , , & \pi_{ij}^{\sf,\ \hat{\!\ti f}\,\ ! ' } & = & \pi_{ij}^{\sf,\ , \ \!\hat{\!\!\sf}\,\ ! ' } ( \hat f ' \rightarrow \hat f')\ , , \end{array}\end{aligned}\ ] ] where the sub/superscript @xmath422 denotes values belonging to first and second generation scalar fermions with same isospin as @xmath25 ( e. g. @xmath423 for the stop case , ) , @xmath424 sfermions with different isospin etc .
@xmath425 a_0 ( m_{h_k^0}^2 ) \\[1 mm ] \non & & \hspace{-3mm}+\frac{1}{(4\pi)^2 } \sum_{k=1}^2 \bigg [ h_{f'}^2
\,d^{\sf'}_{kk}\ , r^\sf_{i1 } r^\sf_{j1 } + h_{f}^2 \,d^{\sf}_{kk}\ , r^\sf_{i2 } r^\sf_{j2 } + g^2 \,f^\sf_{ij}\ , \big ( d^\sb_{kk } - d^\st_{kk } \big ) \bigg ] a_0 ( m_{h_k^+}^2 ) \\ \\
\non \pi_{ij}^{\sf,\,v } & = & \frac{1}{(4\pi)^2 } \,4\ , ( e_0 e_f)^2 { \delta}_{ij } a_0(\l^2 ) + \frac{1}{(4\pi)^2 } \,2\ , g^2 r^\sf_{i1 } r^\sf_{j1 } a_0(m_{\scriptscriptstyle w}^2 ) \\[1 mm ] & & \hspace{-3mm}+\frac{1}{(4\pi)^2 } \,4\ , g_{\scriptscriptstyle z}^2 \bigg [ \big ( c_l^f \big)^2 r^\sf_{i1 } r^\sf_{j1 } + \big ( c_r^f \big)^2 r^\sf_{i2 } r^\sf_{j2 } \bigg ] a_0(m_{\scriptscriptstyle z}^2 ) \end{aligned}\ ] ] h. baer , d. dicus , m. drees , and x. tata , phys .
d * 36 * ( 1987 ) 1363 ; + j. f. gunion and h. e. haber , nucl .
b * 307 * ( 1988 ) 445 ; + j. f. gunion and h. e. haber , nucl .
b * 402 * ( 1993 ) 569(e ) ; + k. griest and h. e. haber , phys .
d * 37 * ( 1988 ) 719 .
a. djouadi , j. kalinowski , and p. m. zerwas , z. phys .
c * 57 * ( 1993 ) 569 ; + a. djouadi , p. janot , j. kalinowski , and p. m. zerwas , phys . lett .
b * 376 * ( 1996 ) 220 ; + a. djouadi , j. kalinowski , p. ohmann , and p. m. zerwas , z. phys .
c * 74 * ( 1997 ) 93 ; + a. djouadi , mod .
lett . a * 14 * ( 1999 ) 359 ; + g. blanger , f. boudjema , f. donato , r. godbole , and s. rosier - lees , nucl .
b * 581 * ( 2000 ) 3 .
r. hempfling , phys .
d * 49 * ( 1994 ) 6168 ; + l.j .
hall , r. rattazzi , and u. sarid , phys .
d * 50 * ( 1994 ) 7048 ; + m. carena , m. olechowski , s. pokorski , and c.e.m .
wagner , nucl .
b * 426 * ( 1994 ) 269 ; + d.m .
pierce , j.a .
bagger , k. matchev , and r. zhang , nucl .
b * 491 * ( 1997 ) 3 .
gorishny , a.l .
kataev , s.a .
larin , and l.r .
surguladze , mod .
a5 * ( 1990 ) 2703 ; phys .
d * 43 * ( 1991 ) 1633 ; + a. djouadi , m. spira , and p.m. zerwas , z. phys .
c * 70 * ( 1996 ) 427 ; + a. djouadi , j. kalinowski , and m. spira , comput .
* 108 * ( 1998 ) 56 ; + m. spira , fortschr . phys .
* 46 * ( 1998 ) 203 . | we calculate the _ full _ electroweak one loop corrections to the decay of the cp
odd higgs boson @xmath0 into scalar fermions in the minimal supersymmetric extension of the standard model .
for this purpose many parameters of the mssm have to be properly renormalized in the on shell renormalization scheme .
we have also included the susy qcd corrections . for the decay into bottom squarks and tau sleptons , especially for large @xmath1
, the corrections can be very large making the perturbation expansion unreliable .
we solve this problem by an appropriate definition of the tree
level coupling in terms of running fermion masses and running trilinear couplings @xmath2 .
we also discuss the decay of heavy scalar fermions into light scalar fermions and @xmath0 .
we find that the corrections can be sizeable and therefore can not be neglected .
hep - ph/0308146 * improved full one loop corrections + to @xmath3 and @xmath4 * + c. weber , h. eberl , w. majerotto + |
in recent years low surface brightness ( lsb ) galaxies have received an increasing amount of attention . with the advent of deep
, large - field surveys many of these galaxies have been discovered , both in clusters and in the field ( schombert et al .
1988 , 1992 [ hereafter sbsm ] , turner et al .
1993 , davies et al .
1994 , schwartzenberg et al .
1995 , sprayberry et al .
1995a ) . the main property which distinguishes lsb galaxies from normal `` classical '' galaxies that define the hubble sequence is their low surface brightness ( @xmath4 ) and not their size .
lsb galaxies are not necessarily dwarfs .
many of them are disk galaxies that typically have scale lengths of several kpc , ( mcgaugh & bothun 1994 , de blok et al .
1995 [ hereafter bhb95 ] ) i.e. , in the same size range as typical high surface brightness ( hsb ) galaxies .
the faintest lsb disk galaxies hitherto discovered have central surface brightnesses of @xmath5 mag arcsec@xmath6 ( turner et al .
1993 ; irwin et al . 1990
; davies , phillips & disney 1988 ) .
this is almost two orders of magnitude fainter than those of the disk galaxies that obey freeman s law ( 1970 ) , which states that most disk galaxies have a central @xmath7-band surface brightness in the range @xmath8 mag arcsec@xmath6 .
the existence of lsb galaxies with central surface brightnesses deviating many @xmath9 from the freeman value indicates that this law is not a complete description of the central surface brightness distribution of disk galaxies .
attempts to construct the proper distribution from extant data ( mcgaugh , bothun & schombert 1995 , mcgaugh 1996 ) suggest that lsb galaxies are numerically common , though considerable uncertainty remains .
de jong ( 1995 , 1996 ) investigated the central surface brightness distribution of spiral galaxies for a range of hubble types .
he concluded , on the basis of a diameter - limited , statistically complete sample of 86 face - on spiral galaxies from the ugc ( including a few lsb galaxies ) , that freeman s law should be restated as that there is no preferred single value for the central surface brightness of disks in galaxies , but only an upper limit in surface brightness which galaxies can have ( corresponding approximately to the freeman value ) ( see also allen & shu , 1979 ) . towards later types the central surface brightness systematically decreases .
in general the range of properties exhibited by lsb galaxies is much larger than that of the classical freeman galaxies . the sizes and luminosities of lsb galaxies range from the very small and faint dwarfs like gr8 ( hodge 1967 ) , to the giant and luminous lsb galaxy malin 1 ( impey & bothun 1989 , sprayberry et al .
lsb galaxies _ in general _ are therefore likely to be not so much a single group , different from all other types of galaxies , but rather the dim counterparts of the known `` classical '' galaxies .
we have drawn our sample from the list of lsb galaxies by sbsm .
the sbsm list was made by scanning by eye selected fields of the new poss - ii survey , and applying the ugc diameter selection criteria to the objects discovered .
the increased depth of these new plates resulted in the discovery of many lsb galaxies that would have been in the ugc had it been compiled at present using the poss - ii plates .
sbsm make no claims about completeness in their survey , but rather that `` our purpose ... is simply to increase our known sample of lsb galaxies and to offset the clear bias in apparent magnitude catalogs for the discovery of only hsb systems . ''
the present paper is written in this spirit
. with this in mind it is striking that lsb galaxy surveys that have concentrated on detecting field lsb galaxies ( like e.g. sbsm ) have in general found that the large majority of the galaxies ( @xmath10 85% in the case of sbsm ) in these surveys are late - type spirals .
ellipticals and early - type spiral galaxies form only a small minority . in our work
we will therefore restrict ourselves to describing the properties of a subset of lsb galaxies : those of the late - type lsb disk galaxies . in the rest of the paper when we refer to lsb galaxies
we will always implicitly mean this sub - group , unless noted otherwise .
not included are dwarf ellipticals and spheroids , like those found in the local group .
the group of late - type lsb galaxies is in many ways an extension of the hubble - type sequence towards very late types .
they continue ( and are in many cases the most extreme representatives known of ) the trends defined by the hsb galaxies along the hubble sequence towards lower surface brightnesses .
such trends are increasingly bluer colours ( bhb95 , mcgaugh & bothun 1994 [ hereafter mb94 ] , rnnback 1993 ) , decreasing oxygen abundances in the gas ( mcgaugh 1994 , rnnback & bergvall 1995 ) , and decreasing hi surface densities ( van der hulst et al.1993 [ hereafter vdh93 ] ) .
a detailed investigation of these late - type lsb disk galaxies will therefore provide an insight into systematic changes of properties of spiral galaxies along the hubble sequence as an explicit function of surface brightness .
one particular example of the trends with surface brightness was presented in zwaan et al ( 1995 ) ( cf .
romanishin et al .
it was shown that late - type lsb disk galaxies also lie on the tully - fisher relation ( see also sprayberry et al .
1995b ) , requiring a systematic and fine - tuned relation between the surface brightness of a galaxy and its overall mass - to - light ratio .
in order to hope to understand this systematic relation , and perhaps to uncover other relations important to our understanding of galaxy properties and their history , resolved hi imaging , providing detailed rotation curves , is required .
we have obtained such hi imaging data for 19 lsb galaxies with the vla and wsrt synthesis radio telescopes . in this paper
we present the data and a global analysis . in forthcoming papers
we will present a full decomposition of the rotation curves into disc and halo components , and , in combination with optical properties , discuss the processes that regulate the rate of evolution and star formation in these dim galaxies . in section 2
we discuss the sample selection .
section 3 contains technical details on the observations . in section 4
we discuss the reduction of the data . in section 5 the derivation of parameters from these data
are given . in section 6
we present the data in tabular and graphical form .
section 7 contains a discussion of the hi surface densities . in section 8
the rotation curves are discussed .
section 9 is devoted to the mass - to - light ratios , and finally in section 10 the main points of this paper are summarized .
we present 21-cm hi synthesis radio observations of a sample of 19 lsb galaxies , taken from the catalogues of lsb galaxies by sbsm .
the sbsm catalogue is not dominated by small , faded dwarf galaxies .
lsb galaxies occupy the full range of types as defined by the hubble sequence , however with a strong preference towards later types .
the redshift distribution is similar to those of the ugc galaxies , with only small deviations at the low and high velocity tails , where the lsb survey picks up lsb dwarfs , and lsb giants respectively .
the spatial distribution shows that lsb galaxies are good tracers of the large - scale structure on scales of @xmath11 mpc ( cf .
bothun et al .
1993 ; mo , mcgaugh & bothun 1994 ) .
they do not fill the voids .
the same applies to the distribution of hi masses : apart from a slightly smaller mean for the lsb galaxies , they are similar .
the large fraction of lsb galaxies that has been detected in hi shows that these galaxies are just as likely to contain hi as their hsb counterparts of similar hubble - type .
we refer to sbsm for a complete discussion of the properties of the galaxies in their catalogue .
the typical lsb galaxy , as listed in sbsm , is therefore a late - type spiral galaxy in our local universe , with luminosity and hi mass not significantly different from late types already present in the ugc , but typically more extreme in surface brightness . in order to select a number of representative galaxies from the sbsm catalogue ,
we have applied the following criteria . 1 .
the galaxy should have been detected in hi .
this necessarily limits the sample to the 171 galaxies that have been observed with arecibo , but as explained in sbsm this does not bias the sample towards objects containing much hi , as the success rate of detections for these galaxies was at least 80% ( and would have been higher if not for the limits imposed by the observable velocity range ) .
2 . as the hi masses of galaxies in sbsm are strongly peaked around @xmath12 with a spread of 0.5 dex to both sides ( cf . fig . 5 in sbsm ) , we have selected only galaxies with hi masses in the range @xmath13 .
this leaves 132 galaxies .
3 . in order not to confuse true dwarfs with lsb disk galaxies
we have furthermore restricted the sample to those galaxies with redshifts between 3000 and 8000 km s@xmath1 , which sbsm define as `` lsb disks '' ( cf .
their fig .
this leaves 95 galaxies .
4 . from this sample of 95 galaxies , we selected 16 galaxies at random .
this subsample represents in our opinion a typical cross - section of the larger sample of late - type lsb disk galaxies .
the sample is listed in tables 13 . in order to compare the properties of these lsb galaxies with those of dwarf lsb galaxies and large lsb galaxies , we further supplemented the sample with examples of these two classes .
the dwarf sample consists of 3 galaxies with hi masses smaller than @xmath14 @xmath15 and redshifts smaller than 3000 km s@xmath1 .
these are f564-v3 , f571-v2 and f583 - 1 .
we used the sample of vdh93 to represent the large lsb galaxies . although their hi masses are slightly larger , they set themselves apart from our main lsb sample mostly because of their larger optical scale lengths , and because of the fact that despite their low central surface brightness , they were included in the ugc catalogue .
although our sample is not complete in a statistical sense , we have obtained a sample of galaxies which is representative of the late - type lsb disk galaxies found in the sbsm catalogue .
we can use this sample to extend the range in surface brightness over which properties of disk galaxies are investigated towards lower surface brightnesses . it should also be noted that , to the best of our knowledge , no hi - rotation curves of galaxies with central surface brightnesses fainter than 23.5 @xmath7-mag arcsec@xmath6 ( the average surface brightness of our sample ) have been published . there are only a handful of rotation curves available in the literature of lsb dwarfs ( @xmath16 ) , most of them ddo dwarfs ( carignan & beaulieu 1988 , hoffman et al .
this is surprising , given the fact that these galaxies are dark - matter dominated , providing ideal cases for the study of the dynamics and density distributions of the dark matter .
the discussion on whether halos are isothermal or singular ( see flores & primack 1994 ) depends very much on evidence provided by this same handful of rotation curves .
the sample presented here thus greatly extends the available data on galaxies in this still ill - charted lsb area of parameter space .
this is currently the largest sample of lsb galaxies for which detailed optical , radio and abundance data are available ( vdh93 , mcgaugh 1994 , mb94 , bhb95 ) .
we observed 19 lsb galaxies in the 21-cm hi line .
observations of 10 galaxies were performed with the vla in its 3-km ( c ) configuration using two if systems , 27 antennas and 64 channels over a 3.125 mhz bandwidth .
the spectra were hanning - smoothed on - line providing a velocity resolution of 20.6 km s@xmath1 .
each galaxy was observed for 5 times 40 minutes interspersed with 5 minute observations of a phase calibrator . for bandpass calibration and the flux density scale we observed 3c286 with the same correlator settings as each galaxy .
the observations were performed on 10 , 16 , 17 and 21 december 1990 .
further details are given in table 1 .
the other 9 galaxies were observed with the wsrt .
each galaxy was observed for 12 hours with 38 baselines ranging from 72 m to 2736 m with 72 m increments . both if systems were used and the correlator was set up to provide 64 channels over a bandwidth of 2.5 mhz .
no on - line spectral smoothing was performed , so the highest possible velocity resolution is 9.9 km s@xmath1 .
off - line hanning smoothing reduces the velocity resolution to 16.5 km s@xmath1 .
amplitude / phase and bandpass calibration was performed using observations of 3c147 and 3c286 .
the observations were done in june and september 1990 and may and august 1991 .
+ _ columns : _ + ( 1),(3 ) name of the galaxy .
+ ( 2),(4 ) source of the data .
+ sources : + [ 1 ] carignan & freeman ( 1985 ) + [ 2 ] carignan & beaulieu ( 1989 ) + [ 3 ] broeils ( 1992 ) + [ 4 ] puche , carignan & wainscoat ( 1991 ) + [ 5 ] carignan & puche ( 1990 ) + [ 6 ] carignan ( 1985 ) + [ 7 ] puche , carignan & bosma ( 1990 ) + [ 8 ] kent ( 1986 ) + [ 9 ] begeman ( 1987 ) + [ 10 ] kent ( 1987 ) + [ 11 ] jobin & carignan ( 1990 ) + [ 12 ] begeman ( 1989 ) + [ 13 ] c^ ot ' e , carignan & sancisi ( 1991 ) + [ 14 ] carignan , sancisi & van albada ( 1988 ) + [ 15 ] roelfsema & allen ( 1985 ) + [ 16 ] bhb95 + [ 17 ] this work + [ 18 ] mb94 + [ 19 ] vdh94 + table a3 contains the identification of the galaxies plotted in fig .
[ large ] . also given
are the sources for the data .
we retrieved the surface brightnesses from the original papers ( see notes to table a3 for a complete list ) , and where necessary converted to the @xmath7-band . | we present very large array ( vla ) and westerbork synthesis radio telescope ( wsrt ) 21-cm hi observations of 19 late - type low surface brightness ( lsb ) galaxies .
our main findings are that these galaxies , as well as having low surface brightnesses , have low hi surface densities , about a factor of @xmath0 lower than in normal late - type galaxies .
we show that lsb galaxies in some respects resemble the outer parts of late - type normal galaxies , but may be less evolved .
lsb galaxies are more gas - rich than their high surface brightness counterparts .
the rotation curves of lsb galaxies rise more slowly than those of hsb galaxies of the same luminosity , with amplitudes between 50 and 120 km s@xmath1 , and are often still increasing at the outermost measured point .
the shape of the rotation curves suggests that lsb galaxies have low matter surface densities .
we use the average total mass surface density of a galaxy as a measure for the evolutionary state , and show that lsb galaxies are among the least compact , least evolved galaxies .
we show that both @xmath2 and @xmath3 depend strongly on central surface brightness , consistent with the surface brightness mass - to - light ratio relation required by the tully - fisher relation .
lsb galaxies are therefore slowly evolving galaxies , and may well be low surface density systems in all respects .
galaxies : evolution galaxies : fundamental parameters galaxies : structure galaxies : spiral galaxies : kinematics and dynamics dark matter |
when batse observes bursts from different red - shifts at a fixed energy band it detects photons from different energy bands at the source . this spectral dependence complicates the interpretation of the peak - flux and time - dilation distributions .
so far several attempts have been made to overcome this problem by modeling the spectral shape of the bursts .
we suggest here a different method which is based on the availability of multi channel data in different energy bands .
the basic idea beyond our scheme is that we
view " all bursts at the same intrinsic energy band independently of their red - shift by by scanning over the different channels until we find the most likely red - shift and using this value to blue - shift " back the observed spectrum to the initial spectrum at the source .
now we look at the same energy band at the source for all bursts , avoiding the issue of the spectral shape .
consider first a population of standard candle sources that have a fixed luminosity , @xmath1 , at a given energy band centered around an energy @xmath2 .
for each burst we have a set of measured peak fluxes @xmath3 at different energy bands .
we determine the red - shift of a particular burst by solving the equation : @xmath4 where @xmath5 is the luminosity distance and @xmath6 is the observed peak - flux at the energy @xmath7 .
one may wonder whether there will be multiple solutions to this equation .
the answer is surprisingly no , provided that the spectrum is softer than @xmath8 . as an example we consider a simulated source with a spectral shape : @xmath9 .
figure 1 shows the luminosity as calculated by this method for simulated sources with @xmath10 ( upper sheet ) and @xmath11 ( lower sheet ) in an @xmath12 cosmology .
the only @xmath13 which solves the luminosity equation is the actual redshift .
once we obtain the red - shift for each burst we proceed to compare the red - shift distribution with the one predicted by a different cosmological models .
this is done , for example , by using the maximum likelihood method .
we check whether the method is consistent by repeating the whole process for different energy channels ( at the source ) .
if there is no noise the different channels should give the same red - shift distribution .
in reality we do nt have standard candle sources .
current estimates suggest that the grb luminosity function is quite narrow ( with variation of peak flux of no more then one order of magnitude ) but it is unlikely that it is a delta function .
thus , the luminosity of a burst is not known a priori , and the redshift can not be deduced uniquely from the peak flux . however
, this method can be used in a statistical manner even when faced with this uncertainty .
we assume that the bursts have a known luminosity function @xmath14 at a standard energy band at the source .
in fact all that we need to assume is that @xmath14 has a given functional shape characterized by a few parameters which will be determined by the analysis . as in the standard candle case
we blue - shift the peak flux in each of the energy channels by a suitable factor @xmath15 to our canonical energy at the source and we calculate the corresponding luminosity : @xmath16 , using equation 1 .
we then estimate the likelihood that the burst is at redshift @xmath13i as : @xmath17 for standard candles @xmath18 and @xmath19 . figure 2 depicts the spread in @xmath13 due to a power - law luminosity distribution : @xmath20 a given cosmological model , with a given evolution model ( that is number density of bursts per co - moving as a function of cosmological time ) determines @xmath21 , the expected number of bursts from a @xmath22 interval centered around @xmath13 , per observer unit time . with a given luminosity function at a fixed energy band at the source
we calculate now the likelihood of a burst , given the set of data @xmath3 : @xmath23
n(z_i ) \ . } \label{like1}\ ] ] once we have this estimate for the likelihood of all bursts we proceed to write the likelihood function of the whole set of data .
then we use the maximal likelihood method to estimate the most likely cosmological parameters and the luminosity function parameters .
it is apparent that as the luminosity distribution widens , the constrain on the redshift weakens , and the information about cosmology is vaguer .
the difference between reasonable spectrums is not large .
for example to confuse a redshift @xmath24 bursts and a redshift @xmath25 with @xmath26 requires a luminosity factor of @xmath27 while @xmath11 requires a luminosity factor of @xmath28 . naturally , as the amount of data increases , the confidence in the determination of @xmath13 increases .
once more , self consistency requires that similar results are obtained when using different energy channels .
the time - dilation analysis should be combined with the peak - flux estimate .
the combined analysis provide a self consistent method which estimates the red - shift of each burst using both sets of data
. this should be more accurate than the standard analysis @xcite , @xcite,@xcite in which the bursts are grouped into large groups of bright , dim and dimmest bursts . using the spectral invariant method we view all bursts in the same intrinsic energy , and in this way we eliminate the energy dependence of the bursts duration .
now we would like to combine information about the duration of a burst in the different energy channels , @xmath29 ( this could be @xmath30 or alternatively the autocorrelation characteristic time ) with the set of peak - fluxes @xmath3 .
we estimate the likelihood of the burst as : @xmath31 dz = \cr \sum_i n(z_i ) \tilde\phi \left [ { c ( { e \over 1+z_i } ) d_l(z_i)^2 ( 1+z_i)^{-2 } ) } \right ] \psi \left [ { { \delta t_i({e \over 1+z_i } ) \over 1+z_i } , e } \right ] } \ ] ] where @xmath32 is the intrinsic energy dependent duration distribution function . using the maximum likelihood method we can now estimate the best - fit parameters of the luminosity function , @xmath33 , the duration distribution , @xmath32 , and the maximal redshift from which the bursts are observed .
the spectrum invariant method is the only way to have a complete usage of all the information available in the multi - spectral channel detectors in situations in which the spectra is being red - shifted differently from burst to burst .
it also uses the time - dilation and the peak - luminosity in a way that more than just brings a consistent answer , it uses the whole data to enlarge the statistical confidence of the results .
a luminosity distribution influences considerably the confidence the red - shift determination of a single burst origin . however , it has a minor influence when using a large number of bursts .
similarly the duration distribution makes it impossible to determine the red - shift of a single burst , however one can determine the parameters of the duration distribution and find whether the red - shift data determined from the time dilation analysis is consistent with the one determined from the peak - flux distribution .
fenimore , e. e. , et al , 1995 , apjl , * 448 * , 101l fenimore , e. e. , and bloom , j. s. , 1995 , apj , * 453 * , 25 .
norris , j. p. , et al , 1994 , apj . * 424 * , 540 . band , d. l. , 1994 , apjl .
wijers , r. a. m. j. , and paczyski , b. , 1994 , apjl , * 437 * , 107l . | the analysis of the batse s count distribution within cosmological models suffers from observational uncertainties due to the variability of the bursts spectra : when batse observes bursts from different redshifts at a fixed energy band it detects photons from different energy bands at the source .
this adds a spectral dependence to the count distribution @xmath0 .
similarly variation of the duration as a function of energy @xcite at the source complicates the time dilation analysis .
it has even been suggested that these methods lead to inconsistent estimates of the redshift from which the bursts are observed@xcite .
clearly it would be best to combine the estimates and to perform a joint analysis of the strength and the duration of the bursts .
but for this we have to eliminate first the spectral dependence problem .
we describe here a new statistical formalism that performs the required blue shifting " of the count number and the burst duration in a statistical manner .
this formalism allows us to perform a combined best fit ( maximal likelihood ) to the count distribution , @xmath0 , and the duration distribution simultaneously .
the outcome of this analysis is a single best fit value for the redshift of the observed bursts . |
the moderate to high redshift universe has been probed at unparallelled depth with the most sensitive observations performed at x - ray wavelengths in the _ chandra _ deep fields .
the _ chandra _
2ms observations ( alexander et al . 2003 , luo et al . 2008 ) resolved about 80 per cent of the extragalactic x - ray light in the hard 2 - 10 kev band ( see brandt & hasinger 2005 for a review ) .
these deep surveys find a sky density of 5000 sources per square degree , the vast majority of which are found to be agn through optical spectroscopy ( e.g. barger et al . 2003 ) .
in contrast , the optical surveys for qsos ( e.g. the combo-17 survey ) reach only a surface density about an order of magnitude lower ( e.g. wolf et al . 2003 ) .
this clearly demonstrates the power of x - ray surveys for detecting agn .
this is because hard x - rays can penetrate large amounts of gas without suffering from significant absorption .
indeed detailed spectral analysis on x - ray selected agn reveals large amount of obscuration ( e.g. akylas et al .
2006 , tozzi et al .
2006 , georgantopoulos et al .
in particular , about two thirds of the x - ray sources , over all luminosities , present column densities higher than @xmath11 .
these high absorbing columns are believed to originate in a molecular torus surrounding the nucleus .
however , even the efficient 2 - 10 kev x - ray surveys may be missing a fraction of highly obscured sources .
this is because at very high obscuring column densities ( @xmath12 , corresponding to an optical reddening of @xmath13 ) , the x - ray photons with energies between 2 and 10 kev are absorbed .
these are the compton - thick agn ( see comastri 2004 for a review ) where the compton scattering on the bound electrons becomes significant . despite the fact that compton - thick agn are abundant in our vicinity ( e.g. ngc1068 , circinus ) , only a few tens of compton - thick sources have been identified from x - ray data ( comastri 2004 ) .
although the population of compton - thick sources remains elusive there is concrete evidence for its presence .
the x - ray background synthesis models can explain the peak of the x - ray background at 30 - 40 kev , where most of its energy density lies , ( frontera et al .
2007 , churazov et al .
2007 ) only by invoking a large number of compton - thick agn ( gilli , comastri & hasinger 2007 ) .
additional evidence for the presence of a compton - thick population comes from the directly measured space density of black holes in the local universe .
it is found that this space density is a factor of two higher than that predicted from the x - ray luminosity function ( marconi et al .
this immediately suggests that the x - ray luminosity function is missing an appreciable number of obscured agn . in recent years
there have been many efforts to uncover heavily obscured and in particular compton - thick agn in the local universe by examining ir or optically selected , [ oiii ] , agn samples .
this is because both the ir and the narrow - line region originate beyond the obscuring region and thus represent an isotropic property of the agn .
risaliti et al . ( 1999 ) examine the x - ray properties of a large sample of [ oiii ] selected seyfert-2 galaxies whose x - ray spectra were available in the literature .
they find a large fraction of compton - thick sources ( over half of their sample ) .
their estimates are complemented by more recent _ _ observations of local agn samples ( cappi et al .
2006 , panessa et al .
2006 , guainazzi et al .
all these authors also claim a large compton - thick agn fraction exceeding 30 per cent of the seyfert-2 population . the advent of the _ swift _ and _ integral _ missions which carry x - ray detectors with imaging capabilities ( e.g. barthelmy et al . 2005 , ubertini et al . 2003 ) in ultra - hard x - rays ( 15 - 200 kev ) try to shed new light on the absorption properties of agn in the local universe . in principle , at these ultra - hard x - rays obscuration should play a negligible role , at least up to column densities as high as @xmath14 .
however , because of the limited effective area the above surveys can provide x - ray samples , down to very bright fluxes @xmath15 , with limited quality spectra . again observations are often required to determine the column density in each source .
interestingly , these surveys find only a limited number of compton - thick sources ( markwardt et al .
2005 , bassani et al .
2006 , malizia et al .
2007 , ajello 2008 , winter et al .
2008 , tueller et al .
2008 , sazonov et al .
2008 ) . here , we present observations of _ all _ 38 seyfert galaxies in the palomar spectroscopic sample of nearby galaxies ( ho et al .
this is the largest complete optically selected agn sample in the local universe analyzed so far .
23 of the seyfert galaxies presented here have already been discussed in previous works ( e.g. cappi et al .
2006 ) . for 5 of them newer observations
are available and are presented here .
the current work should provide the most unbiased census of the agn column density distribution at low redshifts and luminosities .
[ thesample ] the seyfert sample used in this study is derived from the palomar optical spectroscopic survey of nearby galaxies ( ho , fillipenko , & sargent 1995 ) .
this survey has taken high quality spectra of 486 bright ( @xmath16 mag ) , northern ( @xmath17 ) galaxies selected from the revised shapley - ames catalogue of bright galaxies ( rsac , sandage & tammann 1979 ) and produced a comprehensive and homogeneous catalogue of nearby seyfert galaxies .
the catalogue is 100% complete to @xmath18 mag and 80% complete to @xmath16 mag ( sandage , temmann & yahil 1981 ) . for the purpose of this work
we consider all the seyfert galaxies from the palomar survey .
sources lying in - between the seyfert - liner or the seyfert - transient boundary have been excluded .
furthermore seven seyfert galaxies ( i.e. ngc1068 , ngc1358 , ngc1667 , ngc2639 , ngc3185 , ngc4235 , ngc5548 ) , which have been included in the palomar survey for various reasons ( see ho et al .
1995 ) , even though they did not satisfy the survey selection criteria , are also excluded .
there are 40 seyfert galaxies comprising the optical sample .
9 sources are classified as type-1 ( contains types 1 , 1.2 , 1.5 ) and 31 as type-2 ( contains types 1.8,1.9,2 ) seyfert galaxies .
however ngc4051 , ngc4395 and ngc4639 which have been initially classified as seyfert 1.2 , 1.8 and 1 by ho et al .
( 1997 ) has been re - classified as type-1.5 , 1 and 1.5 respectively ( see cappi et al .
2006 , panessa et al .
2006 , baskin & laor 2008 ) .
moreover ngc185 which is classified as a seyfert-2 may not contain an active nucleus since it presents line intensity ratios possibly produced by stellar processes ( ho & ulvestad 2001 ) .
the main characteristics of these sources , taken from ho et al .
( 1997 ) , are listed in table [ optical ] .
some galaxies listed here present @xmath19 fainter than the formal limit of the palomar survey . according to ho et al .
( 1995 ) this discrepancy can be attributed to errors in the apparent magnitudes given in the rsac . [ cols="^,^,^,^,^,^ " , ] column 1 : name column 2 : black hole mass in units of solar masses column 3 : bolometric luminosity in units of ergs s@xmath20 column 4 : accretion rate @xmath21
in this work we present observations of all the seyfert galaxies from the palomar survey ( ho et al . 1995 ) . we find that @xmath2250 per cent of the seyfert population is absorbed by @xmath23 @xmath24 . in this sample
we have identified 3 compton - thick sources which translates to a fraction of @xmath228 per cent .
five more sources possibly host a highly absorbed or a compton thick nucleus . in the very extreme , and rather unlikely case were all these candidates are true compton - thick sources their fraction reaches 20 per cent of the total population .
cappi et al .
( 2006 ) and pannesa et al .
( 2006 ) , also using data from the palomar survey , provide estimates for the fraction of obscured agn in the local universe .
these authors find that about 50% of their sources are obscured ( @xmath5 ) .
their estimates on the fraction of compton thick sources suggest an absolute minimum of 20 per cent of the total population .
this result comes in contradiction with our findings .
however their sample includes 2 objects not fulfilling the palomar survey selection criteria ( see also section [ thesample ] ) .
these are the 2 compton - thick agn ngc1068 and ngc3185 .
when we exclude these an agreement is found .
risaliti et al .
( 1999 ) study the x - ray absorption in a sample of 45 seyfert-2 galaxies finding that a considerable fraction of these are associated with compton - thick nuclei .
a direct comparison with our results is not straightforward since these authors exclude all the sources with @xmath25}>4\times 10^{-13}$ ] .
however we think that only a luminosity cutoff could reveal column density distribution of the population that contributes to the xrb ( see section 5.2 ) .
the xrb synthesis models can provide tight constraints on the number density of compton - thick sources .
these models attempt to fit the spectrum of the x - ray background roughly in the 1 - 100 kev range .
it is well established that a large number of compton - thick sources is needed ( gilli et al .
2007 ) to reproduce the hump of the x - ray background spectrum at 30 - 40 kev where most of its energy density lies ( churazov et al .
2007 , frontera et al . 2007 ) . here
, we compare the fraction of the compton - thick sources predicted by the model of gilli et al .
2007 with our results .
we use the publicly available pompa software.gilli/counts.html ] this predicts the number counts at a given redshift , flux and luminosity range using the best - fit results for the fraction of obscured , compton - thick sources , of the x - ray background synthesis model of gilli et al .
( 2007 ) .
we restrict the comparison to low redshifts @xmath26 and x - ray luminosities in the bin @xmath27 .
as our sample is not flux limited in the x - rays , we have to choose a flux limit deep enough to ensure that all sources in this luminosity and redshift bin are detected .
a flux limit of @xmath28 satisfies this constraint .
the x - ray background synthesis models predict a fraction of compton - thick sources of about 40 % which is higher compared with our results @xmath29% .
only if all the low @xmath30}$ ] seyfert-2 galaxies are associated with a compton - thick nuclei the discrepancy would become less pronounced .
optically selected samples can still miss a fraction of compton - thick agn .
for example , ngc6240 is classified as a liner in the optical while _ bepposax _
observations show the presence of a compton - thick nucleus ( vignati et al .
moreover _ suzaku _ observations ( ueda et al . 2007 and comastri et al . 2007 ) have demonstrated that a small fraction of agn may have a 4@xmath31 coverage , instead of the usually assumed toroidal structure .
these sources will not exhibit the usual high excitation narrow emission lines and therefore will not be classified as agn on the basis of their optical spectrum .
recent results based on _ integral _ and _ swift _ observations reveal a small fraction of compton - thick sources ( e.g. sazonov et al .
2008 , sazonov et al .
2007 , ajello et al . 2008 ) .
in particular at the flux limit of @xmath32 in the 17 - 60 kev energy band , observations find @xmath33 compton - thick sources .
swift_/bat hard x - ray survey failed to identify any compton - thick agn .
this non detection discards the hypothesis that their fraction accounts for the 20 per cent of the total agn at @xmath34 confidence level .
it is true however that some heavily obscured compton - thick sources with @xmath35 would be missed even by these ultra hard x - ray surveys . in the low - luminosity sub - sample ( intrinsic @xmath36 erg s@xmath20 ) the fraction of obscured sources diminishes to 30% .
this result comes in apparent contradiction with recent findings suggesting an increasing fraction of obscuration with decreasing luminosity ( e.g. akylas et al .
2006 , la franca et al . 2005 ) .
this behaviour may reflect a physical dependence of the column density with intrinsic luminosity as suggested by elitzur & shlosman ( 2006 ) .
these authors present a model where the torus and the blr disappear when the bolometric luminosity decreases below @xmath37 ergs s@xmath20 because the accretion onto the central black hole can no longer sustain the required cloud outflow rate .
it is interesting to note that the corresponding luminosity in the 2 - 10 kev band is about several @xmath38 , assuming the spectral energy distribution of elvis et al .
interestingly , almost all of our seyfert-2 sources with no absorption present luminosities below this limit ( with the exception of ngc3147 ) .
we note however , there are sources ( ngc3486 , ngc3982 ) with low luminosity , which present column densities around @xmath11-@xmath39 @xmath24 .
alternatively , it is possible that at least in a few cases , the large xmm - newton point spread function results in contamination by nearby sources
. thus the nuclear x - ray emission could be out - shined giving the impression that there is no obscuration ( e.g. brightman & nandra 2008 ) .
however , both the inspection of the images as well as the stacked spectrum of the unabsorbed sources do not favour such a scenario .
observations are available for all 38 seyfert galaxies from the palomar spectroscopic sample of galaxies of ho et al .
( 1995 , 1997 ) .
our goal is to determine the distribution of the x - ray absorption in the local universe through x - ray spectroscopy .
our sample consists of 30 seyfert-2 and 8 seyfert-1 galaxies .
the results can be summarised as follows : * we find a high fraction of obscured sources ( @xmath5 ) of about 50 % . * a number of sources present low @xmath30}$ ] ratio .
their individual spectra show no evidence of high absorbing column densities .
however , their stacked spectrum shows significant amount of absorption ( @xmath40 ) * considering only the bright sub - sample ( @xmath41 ) , i.e. only these sources which contribute a significant amount to the x - ray background flux , we find that 75 % of our sources are obscured . * in the bright sub - sample
there are at least 3 compton - thick agn translating to a fraction of 15% which is lower than the predictions of the x - ray background synthesis models at this luminosity and redshift range . only if we consider , the rather unlikely scenario , where all seyfert-2 galaxies with a low @xmath30}$ ] ratio are associated with compton - thick sources we would alleviate this discrepancy . * we find a large number of unobscured seyfert-2 galaxies .
all these have low luminosities @xmath42 .
inspection of the images , where available , demonstrates that in most cases these are not contaminated by nearby sources .
furthermore , their stacked spectrum reveals no absorption .
it is most likely that these are genuinely unobscured sources in accordance with the predictions of the models of elitzur & sloshman ( 2006 ) .
references ajello m. , rau a. , greiner j. et al .
, 2008 , apj , 673 , 96 akylas a. , georgantopoulos i. , georgakakis a. , kitsionas s. , hatziminaoglou e. , 2006 , a&a , 459 , 693 alexander d. m. , bauer f. e. , brandt w. n. , garmire g. p. , hornschemeier a. e. , schneider d. p. , vignali c. , 2003 , an , 324 , 8 awaki h. , ueno s. , taniguchi y. , 2000 , adspr , 25 , 797 barger a. j. , cowie l. l.,capak , d. m. , et al . , 2003 ,
aj , 126 , 632 barthelmy s. d. , barbier l. m. , cummings j. r. , et al . , 2005 , ssrv , 120 , 143 baskin a. , laor a , 2008 , apj , 682 , 110 bassani l. , dadina m. , maiolino r. , salvati m. , risaliti g. , della ceca r. , matt g. , zamorani g. , 1999 , apjs , 121 , 473 bassani l. , molina m. , malizia a. , et al .
, 2006 , apj , 636 , l65 bianchi s. , corral a. , panessa f. , barcons x. , matt g. , bassani l. , carrera f. j. , jim@xmath43nez - bail@xmath44n e. , 2008 , mnras , 385 , 195 brandt w. n. , hasinger g. , 2005 , ara&a , 43 , 827 brightman m. , nandra k. , 2008 , mnras , 390 , 1241 cappi m. , panessa f. , bassani l. , et al . , 2006 , a&a , 446 , 459 churazov e. , sunyaev r. , revnivtsev m. , et al . , 2007 , a&a , 467 , 529 comastri a. , 2004 , assl , 308 , 245 comastri a. , setti g. , zamorani g. , hasinger g. , 1995 , a&a , 296 , 1 dickey j. m. , lockman f. j. , 1990 , ara&a , 28 , 215 elitzur m. , shlosman i. , 2006 , apj , 648 , l101 elvis m. , wilkes b. j. , mcdowell j. c. , et al . , 1994 , apjs , 95 , 1 foschini l. , di cocco g. , ho l. c. , et al . , 2002 , a&a , 392 , 817 frontera f. , orlandini m. , landi r. , et al . , 2007 , apj , 666 , 86 georgantopoulos i. , zezas a. , 2003 , apj , 594 , 704 georgantopoulos i. , georgakakis a. , akylas a. 2007 , a&a , 466 , 823 ghisellini g. , george i. m. , fabian a. c. , & done c. 1991 , mnras,248 , 14 gilli r. , comastri a. , hasinger g. , 2007 , a&a , 463 , 79 gliozzi m. , sambruna r. m. , foschini l. , 2007 , apj , 662 , 878 guainazzi m. , matt g. , perola g. c. , 2005 , a&a , 444 , 119 heckman t. m. , ptak a. , hornschemeier a. , kauffmann g. , 2005 , apj , 634 , 161 ho l. c. , 2008 , ara&a , 46 , 475 ho l. c. , ulvestad j. s. , 2001 , apjs , 133 , 77 ho l. c. , filippenko a. v. , & sargent w. l.w .
1997 , apjs,112 , 315 ho l. c. , filippenko a. v. , & sargent w. l. 1995 , apjs , 98 , 477 la franca f. , fiore f. , comastri a. , et al . , 2005 , apj , 635 , 864 leahy d. a. , creighton j. , 1993 , mnras , 263 , 314 luo b. , bauer f. e. , brandt w. n. , et al . , 2008 , apjs , 179 , 19l maiolino r. , salvati m. , bassani l. , dadina m. , della ceca r. , matt g. , risaliti g. , zamorani g. , 1998 , a&a , 338 , 781 magdziarz p. , zdziarski a. a. , 1995 , mnras , 273 , 837 malizia a. , landi r. , bassani l. , et al . , 2007 , apj , 668 , 81 marconi a. , risaliti g. , gilli r. , hunt l. k. , maiolino r. , salvati m. , 2004 , mnras , 351 , 169 markwardt c. b. , barbier l. , barthelmy s. , et al . , 2005 , apj , 37 , 1222 matt g. , fabian a. c. , guainazzi m. , iwasawa k. , bassani l. , malaguti g. , 2000 , mnras , 318 , 173 mcelroy d. b. , 1995 , apjs , 100 , 105 nicastro f. , 2000 , apj , 530 , l65 nicastro f. , martocchia a. , matt g. , 2003 , apj , 589 , l13 panessa f. , bassani l. , cappi m. , dadina m. , barcons x. , carrera f. j. , ho l. c. , iwasawa k. , 2006 , a&a , 455 , 173 panessa f. , bassani l. , 2002 , a&a , 394 , 435 pappa a. , georgantopoulos i. , stewart g. c. , zezas a. l. , 2001 , mnras , 326 , 995 risaliti g. , maiolino r. , salvati m. , 1999 , apj , 522 , 157 sandage a. , tammann g. a. , & yahil , a. 1979 , apj , 232 , 352 sazonov s. , revnivtsev m. , burenin r. , churazov e. , sunyaev r. , forman w. r. , murray s. s. , 2008 , a&a , 487 , 509 sazonov s. , revnivtsev m. , krivonos r. , churazov e. , sunyaev r. , 2007 , a&a , 462 , 57 strder l. , briel u. , dennerl k. , et al .
2001 , a&a 365 , l18 tozzi p. , gilli r. , mainieri v. , et al . , 2006 , a&a , 451 , 457 tueller j. , mushotzky r. f. , barthelmy s. , cannizzo j. k. , gehrels n. , markwardt c. b. , skinner g. k. , winter l. m. , 2008 , apj , 681 , 113 turner m. j. l. , abbey a. , arnaud m. , turner m. j. l. , et al . 2001 , a&a 365 , l27 ubertini p. , lebrun f. , di cocco g. , et al . , 2003 ,
a&a , 411 , l131 ueda y. , eguchi s. , terashima y. , et al .
, 2007 , apj , 664 , 79 vignati p. , molendi s. , matt , g. , et al . , 1999 , a&a , 349 , 57 weaver k. a. , nousek j. , yaqoob t. , mushotzky r. f. , makino f. , & otani c. 1996 , apj , 458 , 160 winter l. m. , mushotzky r. f. , tueller j. , markwardt c. , 2008 , apj , 674 , 686 wolf c. , wisotzki l. , borch a. , dye s. , kleinheinrich m. , meisenheimer k.,2003 , a&a , 408 , 499 | we present _ xmm - newton _ spectral analysis of all 38 seyfert galaxies from the palomar spectroscopic sample of galaxies . these are found at distances of up to 67 mpc and
cover the absorbed 2 - 10 kev luminosity range @xmath0 .
our aim is to determine the distribution of the x - ray absorption in the local universe .
three of these are compton - thick with column densities just above @xmath1 and high equivalent width fek@xmath2 lines ( @xmath3 ev ) .
five more sources have low values of the x - ray to [ oiii ] flux ratio suggesting that they could be associated with obscured nuclei .
their individual spectra show neither high absorbing columns nor flat spectral indices .
however , their stacked spectrum reveals an absorbing column density of @xmath4 .
therefore the fraction of absorbed sources ( @xmath5 ) could be as high as @xmath6 % .
a number of seyfert-2 appear to host unabsorbed nuclei .
these are associated with low - luminosity sources @xmath7 .
their stacked spectrum again shows no absorption while inspection of the images , where available , shows that contamination from nearby sources does not affect the _ xmm - newton _ spectra in most cases .
nevertheless , such low luminosity sources are not contributing significantly to the x - ray background flux .
when we consider only the brighter , @xmath8 , 21 sources , we find that the fraction of absorbed nuclei rises to @xmath9 % while that of compton - thick sources to 15 - 20% . the fraction of compton - thick agn is lower than that predicted by the x - ray background synthesis model in the same luminosity and redshift range
. 2@xmath10 |
in the last few decades there was a considerable development in the theoretical description and experimental measurement of differential cross sections for charged particle impact ionization of atoms and molecules @xcite .
nowadays , the interest is focused on the in - detail analysis of the electron ejection from atomic or molecular targets @xcite .
these analyzes may be performed by measuring and calculating fully differential cross sections which gives us the most complete information about an ionization process .
these quantities describe the entire energy and angular distribution of the ionized electron , residual ion and projectile . previously , based on the semiclassical impact parameter method
we have constructed a theoretical model to calculate fully differential cross sections for single ionization of light atoms .
this model takes into account the projectile
target nucleus interaction classically .
the method was tested in case of the single ionization of helium produced by c@xmath1 ion projectile with an energy of @xmath2 mev / u and good agreement with the experiments was achieved in the scattering plane , while in the perpendicular plane a structure similar to that observed experimentally was obtained @xcite .
more complex and interesting features appear when the target is a molecule . due to the multicenter character of the target , interference patterns occur in the ejected electron spectra .
this phenomenon was observed in the double differential cross section ( ddcs ) data for the ionization of hydrogen molecule by fast ions by stolterfoht @xcite and has been analyzed theoretically by several groups @xcite .
recently , fully differential cross sections were measured and calculated for the ionization of h@xmath0 by both fast and slow electrons @xcite , and interference effects were analyzed in these data series . however , while several attempts have been made in the description of the interference effects in the double differential cross sections , and these effects are relatively well understood @xcite , it was shown recently @xcite , that an accurate description of the role of the projectile target nucleus interaction remains a major challenge to theory .
there are even much more details to clarify for these effects , if they appear in the total differential cross section .
interference effects due to indistinguishable diffraction of the incoming projectile from the two atomic centers were identified in the ddcs as a function of scattering angle for fixed ejected electron energy @xcite .
it was shown , that these kinds of interference structures can be more pronounced , that those in the ejected electron spectrum .
the goal of the present paper is to adopt our previous semiclassical method @xcite to calculate fully differential cross sections ( fdcs ) for the ionization of the hydrogen molecule , and to evidence the interference effects due to the two - center character of the target . in this model ,
the description of the molecule is similar to that from paper @xcite , where total cross sections are calculated and are compared to the experimental data . due to the limitations of a semiclassical model ,
we study the ionization induced by fast charged projectiles , and analyze the interference structures only in the ejected electron spectra . as a test case , the single ionization of the hydrogen molecule by 4 kev electron and 6 mev proton projectiles is considered .
fdcs are calculated and compared in the scattering plane with the experimental data of cherid @xcite and dimopoulou @xcite .
results in the perpendicular plane are also presented . in order to evidence the interference effects due to the two - center character of the target ,
the fully differential ionization cross sections for the h@xmath0 molecule and h atom are compared .
in the semiclassical approximation the projectile is treated separately and it moves along a classical trajectory .
this implies that only the electron system needs to be described by a time - dependent schrdinger equation , while the projectile follows the classical laws of motion . in order to study the ionization process of a small molecule induced by fast charged projectiles ,
first the ionization amplitudes have to be calculated .
as described in @xcite , the first - order transition amplitude for a projectile with impact parameter @xmath3 , velocity @xmath4 and charge @xmath5 , and a certain orientation of the molecular axis @xmath6 may be written as @xmath7 where @xmath8 is the position vector of the active electron and @xmath9 is the position of the projectile along the trajectory with @xmath10 . here
the origin is considered in the center of the molecule .
the @xmath11 is the energy transfer to the active electron , where @xmath12 stands for the ionization energy .
the initial and final state of the active electron is denoted by @xmath13 and @xmath14 , with @xmath15 its momentum vector . in the calculations the coulomb interaction @xmath16 is expanded into a multipole series .
the transition probability for a given impact parameter and orientation of the molecular axis is the square of the transition amplitude modulus @xmath17 because in experiments the molecular orientation is usually unknown , the transition probabilities are averaged over molecular orientation @xmath18 finally , the triple differential cross section ( tdcs ) for the electron ejected in the energy range @xmath19 $ ] and into the solid angle @xmath20 and the projectile scattered into the solid angle @xmath21 can be expressed as @xmath22 because some experimental data @xcite are differential relative to the perpendicular momentum transfer @xmath23 , we calculate also a fdcs expressed in terms of @xmath23 and not of the scattering angle @xmath24 an important part of the model is to assign impact parameter values to certain projectile scattering angles .
as described in detail in @xcite , this may be achieved using the transverse momentum balance @xcite , which states that the momentum transfer is the sum of the transverse components of the electron s and residual ion s momenta .
further it is assumed , that the impact parameter is related to the momentum transfer to the residual ion , and we take into account the projectile - electron interaction separately .
thus , in this model , the projectile target nucleus interaction is accounted for . by this point of view two
main collision types are possible : ( a ) binary collision where most of the momentum transfer is taken by the electron and ( b ) recoil collision where most of the momentum transfer is taken by the target nucleus .
accordingly , two different impact parameter values have to be used in these collision regimes . the transition between these impact parameters
is realized smoothly in the transition regions .
the projectile scattering is treated as a classical potential scattering problem in the field of the target system with nuclear charge @xmath25 @xcite . the simplest way to include
the effect of the electrons around the target nucleus is to consider the potential to be a product of the coulomb potential and the bohr - type screening function @xcite .
here it has to be mentioned , that using this potential we assume that the projectile scattering is produced by a spherical potential .
this implies that in the description of the projectile scattering the two - center nature of the molecule is neglected .
while in the analysis of the cross section as a function of the projectile scattering angle the correct description of the projectile
target nucleus interaction is important @xcite , we assume that the present approximation does not influence significantly the character of our results as a function of electron ejection angle . in the particular case of the hydrogen molecule the initial state is represented by a heitler - london type molecular wavefunction . written only for the active electron @xmath26 where @xmath27 is the effective charge , @xmath28 is the vector associated to the internuclear distance and @xmath29 denotes a normalization factor . in order to separate the dependence on the direction @xmath30 of the molecular axis ,
this is expanded into a legendre series @xcite . in this series
each term is characterized by the quantum numbers @xmath31 . in the followings , the expansion coefficients depending only on the molecular distances
will be denoted by @xmath32 .
the @xmath14 continuum wavefunction is expanded into partial wave series , depending on angular momenta @xmath33 and phaseshift @xmath34 .
the radial part @xmath35 is a wavefunction of the continuum electron moving in the mean field of the residual h@xmath36 molecular ion approximated by the potential @xmath37 this wavefunction is calculated numerically . by these considerations
the ionization amplitude will be @xcite @xmath38 where @xmath39 and @xmath40 are the normalization factors of the initial and final state target wavefunctions , respectively .
the quantum numbers @xmath41 corresponds to the multipole expansion of the coulomb interaction and @xmath42 is the integral over the projectile trajectory . using this expression ,
the transition probability is calculated and averaged over molecular orientation .
the fdcs for ionization of the hydrogen molecule will look as @xmath43 where @xmath44 is a constant depending on projectile charge , velocity and scattering angle , target charge and ejected electron momentum and it differs for the different kinds of fdcs mentioned earlier ( see equations ( [ fdcs3 ] ) and ( [ fdcs5 ] ) ) . while our results will be represented scaled to unity at their highest value in the scattering plane , in our following discussion the @xmath44 constant will play the role of a simple scaling constant .
in order to test the validity of the results of the semiclassical model for the h@xmath0 ionization , calculated fdcs for fast electron and proton projectile impact are compared to the available experimental data @xcite .
[ fig1 ] shows fully differential cross sections for the ionization of the hydrogen molecule by 4.087 kev electron impact .
the projectile scattering angle is @xmath45 and the ejected electron energy is 20 ev .
the top panel of the figure shows tdcs in the scattering plane , while the bottom panel shows the same data in the perpendicular plane .
experimental and theoretical data are each scaled to unity at their highest value in the scattering plane .
together with the results for the molecule ( continuous line ) , tdcs are shown for the same collision process with atomic hydrogen target ( dashed line ) .
further comment needs the fact that the tdcs in the scattering plane for atomic hydrogen target show shoulder structures at angles around 45@xmath46 and 135@xmath46 .
such structures can not occur for pure first - order ionization from an isotropic 1s state . as presented above , in order to take into account the projectile
target nucleus interaction , a certain impact parameter is assigned for every kinematic condition .
two different impact parameter values are used in the two collision regimes .
the transition between these impact parameters is realized smoothly in the @xmath47 and @xmath48 transition regions .
the shoulders appear due to this impact parameter adaptation mechanism .
= 7.5 cm in the scattering plane the theory reproduces well in shape the binary and the recoil peak structure .
however , the semiclassical first - order results show a shift of the binary peak with some @xmath49 relative to the experimental data , which have the angular resolution of less than 1@xmath46 @xcite .
such peak shifts indicates that the first order approximation may not be sufficiently accurate .
experimental data in the perpendicular plane for this particular process was not found . as in the case of our previous studies of the he ionization @xcite ,
the semiclassical theoretical results show a double - lobe structure similar to that reported in the literature for slower collision processes @xcite . in order to evidence the interference effects due to the two - center nature of the target , tdcs for the h@xmath0 molecule and h atom
are compared in fig .
the only difference between these calculations is that the one - center wavefunction describing the h atom is replaced by a two - center wavefunction in case of the molecular target . at the first
look the tdcs distributions for h@xmath0 molecule and h atom are very similar .
however , the recoil peak for the h@xmath0 target is larger than that for the h atom .
the interference effects are analyzed later by the means of the interference factor .
we note here , that we obtain the interference structures by comparing the theoretical results for h@xmath0 and h. because the agreement between our calculations and the experimental data is not perfect , we can not say , that have identified the interference effects in the experimental data .
we have identified the interference effects in the electron ejection spectrum , due to the coherent ejection from the two centers of the molecule .
this effect is obtained by describing the molecular electron by a two - center wavefunction .
we assume , that neglecting the two - center character of the target in the description of the projectile scattering does not influence the obtained interference patterns .
the interference structures in the ejected electron spectrum are analyzed through the interference factor defined by the ratio of the cross section obtained for the molecule and of two independent hydrogen atoms @xcite @xmath50 this interference effect may be detected in the tdcs distribution from non oriented molecules .
the interference factor may be expressed by an analytical approximate formula @xcite @xmath51 where @xmath52 is the momentum imparted to the recoil ion . in fig .
[ fig2 ] the interference factor obtained from our calculations ( continuous line ) is shown for the ionization process discussed earlier and it is compared to the analytical form ( [ analyticali ] ) ( dashed line ) .
the top panel of the figure shows the interference factor in the scattering plane . as expected , it has oscillatory behavior with a strong maximum in the vicinity of the recoil peak .
the semiclassical theory predicts another smaller maximum in the binary peak region , too , suggesting a more complicated behavior than that given by the analytical formula . in the perpendicular plane due to the kinematic conditions the @xmath53 momentum is constant : ( 1 ) the transverse component of the recoil momentum @xmath53 parallel to @xmath54 is constant because the electron momentum is zero in that direction and @xmath54 is fixed ; ( 2 ) the transverse component of @xmath53 perpendicular to @xmath54 is constant because @xmath54 is zero in that direction and the electron energy is fixed .
accordingly , the analytical formula gives a constant interference factor .
in contrast to this result , the semiclassical theory predicts some oscillations symmetric relative to the @xmath55 direction ( bottom panel of the figure ) .
= 7.5 cm tdcs distributions for other projectile energies and scattering angles are analyzed in fig .
the top and middle panels present scattering plane tdcs results for 4.087 kev electron projectile with @xmath56 and @xmath57 scattering angles .
the ionized electron is ejected with 20 ev energy .
the results are in good agreement with experiments both in the shape of the distribution and in the position of binary and recoil peaks . the bottom panel of fig .
[ fig3 ] shows results for 4.167 kev projectile energy and a larger @xmath58 scattering angle .
the electron is ejected with a higher energy , of 100 ev . in this case
the experiments show a narrow binary peak , which is well reproduced by our calculations .
= 7.5 cm fig .
[ fig4 ] shows fdcs results of the semiclassical model for ionization of hydrogen molecule by 6 mev proton projectile impact . in this case
the experimental data of dimopoulou @xcite is differential in ejected electron momentum vector and the perpendicular momentum transfer vector .
the comparison is made for different electron ejection energies and momentum transfers .
on one hand , for larger electron energies , of @xmath59 ev ( top panels of the figure ) the agreement between the experimental data and the present , semiclassical theory is reasonably good .
beside of this , it has to be noted that our first - order results show a shift of the binary peak with some @xmath49 , but because of the experimental errors in the cross section this shift can not be determined precisely .
the cross section ratios for the binary and recoil peaks are also well reproduced .
on the other hand , for lower electron energies , of @xmath60 ev ( bottom panels of the figure ) there is a discrepancy between the experimental and theoretical fdcs distributions . in experiments
the cross section ratio of the binary and recoil peaks is almost 1 .
this feature is not reproduced by the semiclassical model .
the explanation is the same as in the case of cdw - eis model @xcite : the fdcs distribution in the sub - ev region is influenced by the presence of the vibrational autoionizing channel ( not included into the semiclassical model ) which leads to the ejection of very low - energy electrons .
the angular distribution of the autoionized electrons is essentially a dipolar one with respect to the momentum transfer axis .
in conclusion , the theoretical model based on the first order , semiclassical , impact parameter approximation used to calculate fully differential cross sections for single ionization of light atoms has been adopted to calculate fully differential cross sections for the ionization of diatomic molecules .
the method has been tested in case of the ionization of h@xmath0 by 4 kev electron and 6 mev proton projectiles . except for some special cases , the obtained results in the scattering plane show good agreement with the experiments @xcite , and are in agreement with other theories @xcite . the double - lobe structure reported in the perpendicular plane
is also reproduced . however
, for low electron ejection energies the semiclassical model fails to reproduce the experimentally observed more symmetrical electron emission patterns in the scattering plane . in order to analyze the effect of the two scattering centers of the molecule on the cross section ,
we have compared the tdcs for the h@xmath0 with the tdcs obtained for independent atoms .
higher cross sections were obtained for molecules in the region of the recoil peak , and also the width of the binary peak was changed .
these differences may be interpreted as interference effects .
the aim of this work was to test that the semiclassical model , in its simple form , is able to treat complex colliding systems and describe interference patterns in the fully differential cross sections , too .
the present work has been supported by the romanian national plan for research ( pn ii ) , contract no .
10 ehrhardt h , jung k , knoth g and schlemmer p 1986 _ z. phys .
* 1 * 3 ullrich j , moshammer r , drner r , jagutzki o , mergel v , schmidt - bcking h and spielberger l 1997 * 30 * 2917 rder j , erhardt h , pan c , starace a f , bray i and fursa d v 1998 * 31 * l525 drner r , mergel v , jagutzki o , spielberger l , ullrich j , moshammer r and schmidt - bcking h 2000 _ physics reports _ * 330 * 95 ullrich j , moshammer r , dorn a , drner r , schmidt lph , schniidt - bocking h 2003 _ rep . prog . phys . _ * 66 * 1463 schulz m , moshammer r , fischer d , kollmus h , madison d h , jones s and ullrich j 2003 _ nature _ * 422 * 48 and references therein schulz m , moshammer r , fischer d and ullrich j 2004 * 37 * 4055 dimopoulou c 2004 _ phis .
lett _ * 93 * 123203 staicu casagrande e m 2008 * 41 * 025204 baran j l , das s , jrai - szab f , pra k , nagy l , and tanis j a 2008 _ phys . rev .
a _ * 78 * 012710 jrai - szab f , nagy l 2007 * 40 * 4259 jrai - szab f , nagy l 2009 _ nucl .
b _ * 267 * 292 stolterfoht n 2001 _ phys
. lett _ * 87 * 023201 galassi m e , rivarola r d , fainstein p d , stolterfoht n 2002 _ phys . rev .
a _ * 66 * 052705 laurent g , fainstein p d , galassi m e , rivarola r d , adoui l , cassimi a 2002 * 35 * l495 nagy l , kocbach l , pra k and hansen j p 2002 * 35 * l453 stia c r , fojon o a , weck p f , hanssen j , rivarola r d 2003 * 36 * l257 sarkadi l 2003 * 36 * 2153 stia c r , fojn o a , weck p f , hanssen j , joulakian b , and rivarola r d 2002 _ phys
. rev . a _ * 66 * 052709 al - hagan o , kaiser c , madison d , murray a j 2008 _ nature physics _ * 5 * 59 laforge a c , egodapitiya k n , alexander j s , hasan a , ciappina m f , khakoo m a and schulz m 2009 _ phys .
_ * 103 * 053201 alexander j s , laforge a c , hasan a , machavariani z s , ciappina m f , rivarola r d , madison d h and schulz m 2008 _ phys .
a _ * 78 * 060701(r ) nagy l and vgh l 1992 _ phys .
a _ * 46 * 284 - 289 chrid m , lahmam - bennani a , duguet a , zurales r w , lucchese r r , dal cappello m c and dal cappello c 1989 * 22 * 3483 roger g. newton , scattering theory of waves and particles , second edition , mineola , new york ( 2002 ) isbn 0 - 486 - 42535 - 5 everhart e , stone g and carbone r j 1955 _ phys . rev . _ * 99 * 1287 | fully differential cross sections are calculated for the ionization of h@xmath0 by fast charged projectiles using a semiclassical model developed previously for the ionization of atoms . the method is tested in case of 4 kev electron and 6 mev proton projectiles .
the obtained results show good agreement with the available experimental data .
interference effects due to the two - center character of the target are also observed and analyzed . |
the progress in optomechanical systems , where an optical or microwave cavity is coupled to a mechanical resonator , was impressive in recent years@xcite .
the accomplishments in optomechanics include cooling mechanical resonator to its quantum ground state@xcite , prediction@xcite and observation@xcite of the optomechanically - induced transparency , squeezing of the cavity@xcite , and mechanical@xcite modes , and coherent state transfer@xcite .
many of these experiments have been realized using superconducting circuits , which enables to consider microwave cavities coupled to mechanical motion as possible building blocks for quantum information processing@xcite .
the coupling between cavity and mechanical resonator plays a central role in optomechanics . in the published experiments ,
intrinsically weak radiation pressure coupling was amplified by increasing drive power of the cavity , which linearizes the effective optomechanical interaction of the system .
such linear interaction , for example , turns gaussian states of the cavity and mechanical resonator into gaussian states . in order to create more general states for quantum information applications and to achieve , for instance , negative wigner function one needs to use either single - photon sources and photodetectors@xcite or non - linear effects , of which non - linear optomechanical interaction is the most common one .
therefore , having strong single - photon radiation pressure coupling of the order or larger than the cavity decay rate is desirable as well as having strong coupling of the cavity to the position squared of the mechanical resonator@xcite .
if the single - photon radiation pressure coupling can be made of the order of the mechanical frequency and larger than the cavity decay rate , the the system is in the ultra - strong coupling regime and photon blockade can be observed@xcite .
along with the ultracold atoms@xcite , superconducting circuits are promising candidates to reach ultrastrong coupling .
recently , the idea of using josephson effect to enhance optomechanical couplings has been researched theoretically@xcite and experimentally@xcite . many of those proposals involve using superconducting quantum interference device ( squid ) with two josephson junctions , which makes cavity intrinsically nonlinear due to the josephson effect .
squid is either embedded into the resonator itself or squid with embedded mechanical resonator is incorporated into a microwave cavity . in this article
, we consider a squid with two symmetric or asymmetric josephson junctions and an embedded mechanical resonator and show that it by itself can produce ultrastrong optomechanical coupling .
originally , a dc squid with embedded mechanical oscillator was studied as a sensitive displacement detector @xcite , however , the asymmetry of the junsctions so far was not at the focus of attention , and theoretical proposals are routinely assuming that two junctions of the squid are almost identical .
a certain asymmetry is always present in the experiment , and we show that it affects the coupling strength .
in addition , we express the couplings in such squid devices in the language of optomechanics , perform numerical simulations of the coupling rates for realistic experimental geometries . doing so we find that this platform has the potential to reach both the single - photon strong coupling , a regime of strong quadratic coupling of the motion to the cavity , and potentially the ultrastrong coupling regime where the single - photon coupling rate exceeds the mechanical frequency . in the first part of the article , we investigate in details the effect of asymmetry in the squid with two junctions and embedded mechanical resonator . as a first step we look at the most common experimental case of the mechanical frequency being much smaller than the cavity frequency@xcite .
we quantize the asymmetric system to get radiation pressure interaction and cross - kerr type interaction , where the cavity is coupled to the position squared of the mechanical resonator .
we show that for experimentally feasible parameters radiation pressure coupling can reach single - photon strong coupling regime and for stronger magnetic fields the ultrastrong coupling regime .
the cross - kerr coupling is usually smaller than radiation pressure coupling but it is the leading coupling at the extremum points of the flux where the radiation pressure is zero . such strong coupling would enable a quantum non - demolition measurement of a phonon number in the mechanical resonator@xcite or the cavity s photon number . as a second step , we study the case when the mechanical and cavity frequencies are of the same order .
since the squid cavity frequency is measuring in ghz , the same order would be required for the mechanical oscillator .
currently , carbon nanotube ( cnt ) resonator can reach ghz frequency@xcite and , hence , the realizations of the squid with suspended cnt junctions@xcite could reach this regime . in this case
, there are two leading interactions : cross - kerr and single - photon beam splitter .
the single - photon beam splitter exists only at the finite asymmetry .
the radiation pressure term is oscillating too fast and is , therefore , disregarded .
the beam splitter is used in many exrimental setups , and hamiltonian with the beam splitter interaction is easily diagonalized and solved .
when single - photon beam splitter is in the range of the strong coupling , one can observe _
e.g. _ optomechanical normal - mode splitting@xcite .
the remainder of the article is organized as follows . in sec .
[ sec : current ] we find current and cavity frequency of the squid with asymmetric josephson junctions and an embedded mechanical resonator . in sec .
[ sec : quantization ] we derive the effective hamiltonian of this system for two cases . in the first case ,
the cavity frequency of the squid is taken to be much larger than the mechanical frequency , which results in the radiation pressure and cross - kerr interactions . in the second case , the cavity frequency
is considered to be of the order of the mechanical frequency providing single - photon beam splitter and cross - kerr interactions . in sec .
[ sec : dis ] we draw the potential map and discuss optomechanical couplings .
finally , we conclude our results in sec . [
sec : concl ] .
+ + in this section , we follow the standard textbook treatment of the current through an asymmetric squid .
we consider two josephson junctions with different values of critical current @xmath0 and @xmath1 connected in a loop together with an embedded mechanical resonator , as shown in fig .
[ fig : freq_i](a ) .
the energy scales for such squid are described by average josephson energy @xmath2 and charging energy @xmath3 with the shunting capacitance of each junction @xmath4 .
the squid has a loop area @xmath5 with the suspended arm of the length @xmath6 .
oscillations of the mechanical resonator modulate the total flux of the squid loop .
then , the squid with the embedded mechanical resonator can be viewed as an lc circuit , in which the josephson inductance of the squid @xmath7 , which for symmetric junctions @xmath8 is well - known to be @xmath9 , changes with the total flux @xmath10 threading though the loop and , consequently , the mechanical resonator couples inductively to the squid , see fig . [
fig : freq_i](b ) . for simplicity , we assume that mechanical resonator moves in its single mode .
the dynamics of the mechanical resonator is described by the displacement @xmath11 from the equilibrium position .
the dynamics of the squid itself is described by the sum of the gauge - invariant phases across each junction ( @xmath12 and @xmath13 ) , @xmath14 , which is referred as the overall phase of the squid .
moreover , the difference of the phases is bound by the total flux threading the loop , @xmath15 where @xmath16 is an integer and @xmath17 is the flux quantum . assuming the magnetic field is applied under certain angle to the squid loop
the total flux can be separated to two contributions .
the first contribution is a bias flux @xmath18 , which is added to the squid loop .
the second contribution comes from a flux threading through the area described by the oscillations of mechanical resonator .
since it is more convenient to work with the phase difference rather than the flux itself we define renormalized bias flux @xmath19 and renormalized flux shift provided by the resonator @xmath20 with the average geometric constant @xmath21 , which takes into account the direction of the magnetic field and the geometry of the mechanical resonator . then , the phase difference is given by @xmath22 here , we study the situation when the circuit has a negligible self - inductance .
now we write the total current going through the asymmetric squid . for this purpose
we introdice the average critical current @xmath23 .
the critical currents of the first and the second junctions are defined as @xmath24 and @xmath25 , respectively , with the asymmetry parameter @xmath26 . therefore , the total current @xmath27 through both junctions is separated to two terms : one is the same as in the case of equal critical currents and another term , which is responsible for the influence of asymmetry , @xmath28 in order to find a critical current of the asymmetric squid , we shift the position of the overall phase of the squid by the phase @xmath29 which satisfies the relation : @xmath30 . then , the total current is simplified to the following @xmath31 where @xmath32 is a flux dependent function , which turns to cosine at zero asymmetry , and the total current becomes the well - known current of the symmetric squid .
then , when mechanical resonator is at rest , we can define the maximum current and , hence , the critical current of the asymmetric squid as well as the cavity frequency @xmath33 here we use @xmath34 , which is the function of @xmath35 instead of @xmath36 at zero displacement . in fig .
[ fig : freq_i](c ) we show the behavior of the critical current for symmetric and 50% asymmetry cases . for the identical junctions the current changes from 0 to @xmath37 , but in the presence of the asymmetry
the current never reaches zero value . even at half flux quantum
when the critical current for the symmetric case is zero , the critical current of the asymmetric squid is at minimum @xmath38 . nevertheless , the maximum , which happens at the odd integer flux quantum , is not affected by the asymmetry .
the cavity frequency is proportional to @xmath39 and portrays the same behavior of the critical current as shown in fig .
[ fig : freq_i](d ) . for parameters of the critical current of the josephson junction @xmath40 na and capacitance
@xmath41 pf the maximum cavity frequency is 10 ghz . at half flux quantum and 50% asymmetry , the cavity frequency reaches its minimum of 4.5 ghz .
in the following , we quantize the system by starting with the classical hamiltonian , which consists of the simple harmonic oscillator , kinetic energy and the potential energy of the squid , @xmath42 where @xmath43 and @xmath44 are the mass and frequency of the mechanical resonator .
the potential energy of the squid @xmath45 is derived from the total current @xmath46 found in sec . [
sec : current ] , @xmath47.\quad \label{energy}\end{aligned}\ ] ] the minimum of the potential is shifted by the flux dependent parameter , which also depends on the displacement of the mechanical resonator .
depending on the difference between the cavity frequency and the mechanical frequency one can assume quasi - static regime or has to take into account the displacement dependent shift . in the typical case
when the mechanical frequency is much smaller than cavity frequency , the shift by the flux can be assumed static on the timescales related to the squid .
then , we can write potential energy in terms of the shifted phase , @xmath48 . the kinetic energy of the squid is not affected by the constant shift , and thus the phase @xmath49 can be replaced by @xmath50 . in order to quantize the phase and the position ,
the potential energy is expanded in terms of the phase up to the second order .
this means that we consider squid as a linear harmonic oscillator with the single - photon kerr shift smaller than the linewidth of the cavity and the cavity frequency .
the term , which is independent of the phase , shifts the equilibrium position of the mechanical resonator and modifies the mechanical frequency @xmath51 for the phase dependent terms we introduce creation and annihilation operators @xmath52 with the momentum coordinate @xmath53 , where @xmath54 is the mass of the phase , and the displacement dependent cavity frequency is @xmath55 this expression can also be retrieved from eq .
for the mechanical resonator at rest by changing @xmath35 to @xmath36 .
therefore , the displacement dependent cavity frequency as a function of @xmath36 has the same behavior as shown in fig .
[ fig : freq_i](c )
. now our hamiltonian has a similar form to that of the hamiltonian with symmetric josephson junctions except for the modified cavity frequency , @xmath56 the position of the mechanical resonator is quantized by introducing the position operator , which is @xmath57 , where @xmath58 and @xmath59 are creation and annihilation operators and @xmath60 is the amplitude of zero point fluctuations of the displacement @xmath11 .
then , the uncoupled hamiltonian of the mechanical resonator is @xmath61 .
the interaction terms are obtained by expanding the displacement dependent cavity frequency to the second order in displacement .
then , the interaction hamiltonian after applying the rotation - wave approximation becomes @xmath62 where the radiation pressure coupling and cross - kerr coupling between cavity and mechanical resonator are , respectively , @xmath63 note even when @xmath64 , the first term in eq .
( [ crosskerr ] ) stays finite because @xmath65 in the radiation pressure coupling is multiplied with the infinite factor @xmath66 .
+ in order to visualize the resulting couplings , to the chosen capacitance @xmath4 and critical current @xmath67 we add following set of parameters : @xmath68 mhz , @xmath69 , @xmath70 , and @xmath71 pg . the flux bias varies from @xmath72 to @xmath73 , where @xmath74 corresponds to chosen value of magnetic field . in fig .
[ fig : gdiff](a ) , we plot the radiation pressure coupling . for the perfectly symmetric josephson junctions , the absolute value of the radiation pressure infinitely increases while getting closer to the half - integer flux quantum .
it suggests that if in the experiment one can tune bias flux very close to half flux quantum the radiation pressure will be maximum .
however , because of the asymmetry of the squid the maximum of the radiation pressure coupling shifts to the value of the flux given be @xmath75 .
behind this value , the radiation pressure monotonically decreases to zero at the half - integer flux quantum .
the maximum of the radiation pressure even at 50% asymmetry and magnetic field of 10 mt can reach single - photon strong coupling regime , considering a typical cavity decay rate of 80 khz .
since the radiation pressure can also be written in terms of the cavity frequency derivative , we can analyze this coupling looking at fig .
[ fig : freq_i](c ) by changing @xmath35 to @xmath36 as mentioned above . for the asymmetric case the slope of the frequency increases and then decreases while changing flux from 0 to @xmath76 . after crossing @xmath76 to @xmath77
it changes sign of the slope , which leads to the negative radiation pressure .
also , for the asymmetric junctions the slope at integer and half integer flux quantum is zero .
the cross - kerr coupling is shown in fig .
[ fig : gdiff](b ) .
the coupling @xmath78 is overall much weaker than radiation pressure except for the odd integer flux quantum , where @xmath79 is zero and @xmath78 is the leading term . for the symmetric case
the coupling is infinitely strong approaching half - integer flux quantum , which is the same behavior found for the radiation pressure , but in contrast to the latter it does not change the sign while crossing @xmath76 . looking at the fig .
[ fig : freq_i](c ) we expect that for the asymmetric squid the cross - kerr coupling , which is the second derivative of frequency , changes the sign between 0 to @xmath76 and then from @xmath76 to @xmath77 and this is indeed the result observed here .
the maximum of the coupling for the asymmetric case is achieved at the half - integer flux quantum .
we can also notice that even at integer flux quantum the value of the cross - kerr coupling ( for the chosen parameters ) is @xmath80 hz . in the experiment with the membrane inside the cavity@xcite the value of the second derivative of the cavity frequency was @xmath81 khz nm@xmath82 .
then in order to improve this value multiple modes of the cavity were coupled to the single mode of the mechanical resonator@xcite to get @xmath83 mhz nm@xmath82 , which is still smaller than our calculated value at integer flux quantum , which is @xmath84 ghz nm@xmath82 .
the maximum of the asymmetric couplings increases with magnetic field , which is captured in fig .
[ fig : gdiff_vs_b ] .
increasing magnetic field to 1 t is experimentally feasible @xcite and increases chances of getting higher couplings .
the radiation pressure coupling is linearly dependent on @xmath85 and the cross - kerr coupling is quadratically dependent on @xmath85 . at 50% asymmetry and magnetic field of 1 t
the radiation pressure can reach an ulrastrong coupling regime ( @xmath86 ) , which also mean that at lower asymmetry the value on the sweat spot can even be greater .
the cross - kerr coupling can reach values of 80khz .
it can be stronger for the lower asymmetry , but the window to catch sweat spot becomes more narrow for the lower asymmetry . in the case of the resonant cavity frequency and mechanical frequency ,
the minimum of the potential is shifted by the position dependent parameter .
however , the displacement is now one of the dynamical variables of the system , separate from the overall phase . if we shift the phase by the displacement dependent parameter then the kinetic energy acquires the shifted phase as well as extra terms in the form of @xmath87 with the original phase .
therefore , it is simpler to expand the arctangent in the potential energy to the first order in @xmath11 , which is sufficient since the amplitude of the mechanical resonator is usually small in such devices .
the expanded potential energy depends on both the displacement and the phase @xmath50 , which do not combine in a single variable , @xmath88 similarly to the previous case , the first term shifts the equilibrium position of the mechanical resonator and the mechanical frequency , @xmath89 next , we quantize the phase introducing the operators @xmath90 , @xmath91 .
the momentum variable @xmath92 stays the same as in the previous case and the displacement dependent cavity frequency is different from eq.([cavity_adiabatic ] ) , @xmath93 then , the hamiltonian in terms of the cavity operators has the following form @xmath94 we expand the full hamiltonian to the second order in the displacement and use the creation and annihilation operators @xmath95 , @xmath58 of the mechanical resonator .
it leads to the uncoupled cavity hamiltonian @xmath96 and the uncoupled mechanical resonator hamiltonian @xmath97 . applying the rotating - wave approximation results in the interaction hamiltonian of the following form @xmath98 where the cross - kerr and the single - photon beam splitter couplings , respectively , are @xmath99 to plot these couplings , we use set of parameters for the nanosquid with cnt junctions@xcite : @xmath100 na , @xmath101pf , @xmath102 , @xmath103 , and @xmath104 ag .
the cavity frequency for this values is 1 ghz .
the mechanical frequency is taken to be @xmath105 ghz , which is possible to reach with a suspended cnt .
the flux bias varies from @xmath72 to @xmath73 , where @xmath106 . + in fig .
[ fig : gdiffr](a ) , we plot the cross - kerr coupling .
the coupling @xmath78 is overall weaker than the one we found in fig .
[ fig : gdiff](b ) .
for the symmetric case , the behaviour is the same as we have seen in the dispersive regime .
however , for finite asymmetry the behavior is qualitatively different since the coupling is negative and does not change the sign from negative to positive .
also , there is a peak of the coupling close to @xmath76 and at exactly half flux quantum there is a minimum .
this happens because of the second term in eq . , which arise due to the modified cavity frequency . for lower asymmetry the minimum is getting closer to zero while the maximum is increasing .
[ fig : gdiffr](b ) shows the single - photon beam splitter coupling . for @xmath107
this coupling is negligible , because the corresponding term in the hamiltonian is a quickly oscillating function of time .
it only exists at a finite asymmetry .
the single - photon beam splitter coupling is slowly increasing while flux rises from 0 to @xmath76 and reaches its maximum at the half flux quantum and then passing this point decreases again . for lower asymmetry
the peak is higher , but the window to reach higher value is narrower , since the higher asymmetry corresponds to a higher value of the coupling except for close to half - integer flux quantum . in fig .
[ fig : gdiffr_vs_b ] we show the maximum of both couplings at the 50% asymmetry while increasing magnetic field to 1 t. the cross - kerr coupling reaches 1 khz .
the single - photon beam splitter coupling has the value up to 10 mhz . in comparison with the mechanical frequency
it is not in the ultrastrong coupling regime , but depending on the cavity decay rate it can be in the strong - coupling regime .
while often radiation pressure is linearized to produce the beam splitter term to solve specific systems and phenomenon , @xmath108 has intrinsically a beam splitter character even at the single - photon level . at very small asymmetry and
strong magnetic fields the cross - kerr coupling can be larger than beam - splitter interaction .
to gain the intuition about the couplings and to better understand the asymmetric system we plot the potential energy as a function of flux ( @xmath36 ) and phase ( @xmath49 ) in fig .
[ fig : pot_energy ] . for the symmetric junctions
the potential is symmetric along dashed line . at
the bottom of the potential radiation pressure is zero and cross - kerr coupling is finite . for the asymmetric case the potential has an elliptical form and
is asymmetric .
one can further study the figures on the bottom corresponding to the cross - section of the energy map for different flux .
the cross - sections are chosen by moving the horisontal dashed line up . in the symmetric case the minimum of the potential energy stays at the same position .
in contrast , for the asymmetric junctions the position of the minimum shifts , which is also described by eq .. in the case of dispersive frequencies , this shift it constant and the minimum is redefined at each bias flux , which represents the same physics as for the symmetric case .
however , elliptical form alters the radiation pressure and cross - kerr couplings and changes its behavior as previously appeared in fig .
[ fig : gdiff ] around half - integer flux quantum where there is a merge between elliptical forms . for the case of resonant frequencies and asymmetric squid , the minimum of the potential energy is shifted by the displacement dependent flux .
the oscillations of the mechanical resonator correspond to the motion from one curve to another one .
the force that triggers the motion between the minima is just like the lorentz force , which explains the appearance of the single - photon beam splitter and the extra term picked up by cross - kerr coupling as compared to the dispersive regime .
we now discuss the mechanical frequency shifts due to the josephson term in eqs . and .
in the parameter regime we have chosen this shift can be disregarded .
however , for higher josephson energy , which could depend on the magnetic field itself , or larger parameters of the mechanical resonator increasing magnetic field to 1 t can create a large shift , which should be taken into account .
the mechanical zero - point fluctuation in such situation becomes smaller , subsequently the values of the couplings decrease , but we do not expect the change in the overall behavior of the optomechanical couplings . generally , the squid is an intrinsically nonlinear cavity and close to the half integer flux quantum an extra nonlinear kerr - type term @xmath109 appears in the hamiltonian of eqs . and , where the kerr nonlinearity is @xmath110 .
this term results from the expansion of the potential energy to the forth order in the overall phase @xmath50 .
thus , a cavity can be considered linear as long as @xmath111 is less than the cavity linewidth and @xmath112 , which gives a finite condition for the flux bias close to the half integer flux quantum .
close to the half flux quantum the kerr - type term and the cross - kerr term are always present in this system . from the fourth order expansion of the potential energy
there are also other nonlinear interaction terms such as @xmath113 in the dispersive case or @xmath114 in the resonant case , which are always small .
we provided a quantum analysis of the squid with asymmetric josephson junctions and embedded mechanical resonator for two cases of the dispersive and resonant cavity and mechanical frequencies .
our findings are significant for the experimental setup where asymmetry can not be avoided .
we found that the radiation pressure for the resonant frequencies has a sweet spot , which is located at an asymmetry dependent flux point .
shifting this point towards the half - integer flux quantum results in a weaker coupling .
even at 50% asymmetry and weak magnetic field the radiation pressure coupling can be in the strong coupling regime . for high magnetic fields
, the ultrastrong coupling regime of the radiation pressure can be achieved .
the cross - kerr coupling is finite at the odd integer flux quantum , in contrast to the radiation pressure coupling .
for the symmetric case , it is always negative and infinitely strong very close to the half - integer flux quantum . for the dispersive asymmetric case ,
the cross - kerr coupling has maximum at half - integer flux quantum and changes sign from negative to positive while reaching maximum . for the resonant asymmetric case
, the minimum sits at the half - integer flux quantum and maximum is at the flux dependent point close to @xmath76 . for the resonant case ,
the radiation pressure is too weak since it oscillates at higher frequency and instead single - photon beam splitter interaction is the main term in the hamiltonian .
it is always finite and has its maximum at the half - integer flux quantum .
we explained the origin of different couplings using the potential energy map as well as compared the maps for the symmetric and asymmetric cases .
the biggest challenge to experimentally work with single - photon beam splitter coupling is the condition on the mechanical frequency , which should be comparable to the cavity frequency . experiments involved such setup previously@xcite had mechanical frequency smaller than cavity frequency .
however , using carbon nanotubes as a mechanical resonator coupled to the josephson circuit can potentially solve the high mechanical frequency issue .
this work was supported by the netherlands foundation for fundamental research on matter ( fom ) . | we investigate superconducting interference device ( squid ) with two asymmetric josephson junctions coupled to a mechanical resonator embedded in the loop of the squid .
we quantize this system in the case when the frequency of the mechanical resonator is much lower than the cavity frequency of the squid and in the case when they are comparable . in the first case ,
the radiation pressure and cross - kerr type interactions arise and are modified by asymmetry .
cross - kerr type coupling is the leading term at the extremum points where radiation pressure is zero . in the second case ,
the main interaction is single - photon beam splitter , which exists only at finite asymmetry .
another interaction in this regime is of cross - kerr type , which exists at all asymmetries , but generally much weaker than the beam splitter interaction .
increasing magnetic field can substantially enhance optomechanical couplings strength with the potential for the radiation pressure coupling to reach the single - photon strong coupling regime , even the ultrastrong coupling regime , in which the single - photon coupling rate exceeds the mechanical frequency . |
three dimensional topological insulators host unusual surface states that can be described in a number of different ways@xcite . in models of free electrons that respect time reversal and charge conservation symmetries which are necessary to describe this phase , the surface is metallic .
the surface electronic structure is comprised of an odd number of 2d dirac cones , which is impossible to realize in a purely 2d system with time reversal invariance .
it is crucial that electrons transform as kramers pairs i.e. time reversal @xmath0 acting twice on electrons gives @xmath4
. other surface terminations of the topological bulk are also interesting . if @xmath0 is broken only at the surface , the metallic edge can be gapped to yield an insulating surface .
the topological bulk properties are revealed in the properties of domain walls between opposite @xmath0 breaking regions on the surface .
the domain walls are necessarily metallic and host an odd number of chiral dirac fermions @xmath5@xcite .
thus , each domain is associated with hall conductance @xmath6 , ( and @xmath7 in the natural units for thermal hall conductance ) where @xmath8 is an integer .
such a half - integer hall conductance if forbidden in a 2d system in the absence of electron fractionalization and is directly related to the magnetoelectric effect@xcite .
the magnetoelectric polarizability of topological insulators is @xmath3 * ? ? ?
* in contrast to trivial , time reversal symmetric insulators which have @xmath9 .
if , on the other hand we break charge conservation on the surface , by pairing and condensing cooper pairs , the resulting surface superconductor also has exotic properties - its vortices host majorana zero modes@xcite , which only occur in 2d systems if @xmath0 symmetry is broken .
it was believed that these were the only possible surface states of the 3d topological insulator - i.e. they must either be gapless ( e.g. metallic ) or break one of the defining symmetries .
recently , inspired by the study of bosonic topological phases@xcite , and fermionic topological superconductors @xcite , a different surface termination has been recognized , that preserves all symmetries and develops an energy gap at the surface@xcite . in this situation
the surface develops topological order i.e. there are anyonic excitations bound to the surface .
while the topological order itself may be realized in 2d , the transformation properties of anyons under action of the symmetry is unlike in any 2d system .
a close analog of this problem was recently discussed in the context of fermionic topological superconductors protected by @xmath0 symmetry , where a non - abelian surface topological order was identified@xcite .
the discussion in this paper will closely follow this earlier work .
of course , such surface states will only be realized in systems with strong electron correlations - and represents a qualitatively new property of interacting topological insulators .
this is analogous to the argument that in a 2d quantum spin system with s=1/2 per unit cell , a gapped , symmetric state must be topologically ordered@xcite . indeed
the absence of ordering is taken as a sign of topological order . similarly ,
if on the surface of a 3d topological insulator , no sign of superconductivity or @xmath0 breaking is present but an energy gap opens , this will be indicative of topological order .
in fact , as discussed in the context of fermionic topological superconductors@xcite , and as we will see below , the topological order here is required to be _ non - abelian_. this mechanism may provide a route to realizing non - abelian topological order , which would provide further impetus in the search for strongly correlated topological insulators such as smb@xmath10 @xcite .
let us discuss some of the key physical requirements that the topologically ordered surface of a 3d ti should satisfy .
( a ) ` @xmath11 ness ' : given the @xmath11 classification of free fermion tis , a pair of topologically ordered surfaces should ` unwind ' , and should be connected to a trivial confined phase .
( b ) magnetoelectric response @xmath3 : in the absence of topological order , surface domains of the @xmath0 breaking insulating surface can be assigned a hall conductivity @xmath12 and thermal hall conductivity @xmath13 ( modulo integers ) .
these are equivalent to the statement that the magnetoelectric response is @xmath3 .
the implication for the topologically ordered states , is that in their 2d versions , which break @xmath0 , should have @xmath2 and @xmath13 modulo integers .
( c ) on breaking charge conservation symmetry , it should be possible to remove the topological order while preserving @xmath0 .
the resulting superconductor should host hc/2e vortices with majorana zero modes in their core .
at first , a promising choice appears to be the read - moore pfaffian state@xcite .
one picture of this phase is to consider a topological superconductor of spin polarized electrons in a @xmath14 pairing state , while the cooper pairs form a @xmath15 bosonic laughlin state .
the latter has anyon excitations with fractional charge @xmath16 and a charged chiral edge state .
the bound state of the superconductor quasiparticle and the charge @xmath17 excitation is identified as the electron and only those excitations that braid trivially with the electron are retained .
this phase has 12 quasiparticles ( including the electron ) and @xmath18 .
while this satisfies one of our criteria , it is readily seen from the topological spins of the associated topological order that this can not be realized in a @xmath0 invariant way even on a surface .
however , a simple modification produces a more promising candidate which we call the t - pfaffian ( @xmath0 preserving pfaffian ) .
consider the time reversed superconductor ( @xmath19 ) combined in the identical way with the abelian topological order of cooper pairs .
this theory also has @xmath2 and @xmath13 when realized in 2d as required .
moreover , the topological spins of the quasiparticles ( table [ t_spin ] ) now appear compatible with time reversal symmetry .
we construct an exactly soluble 3d model that explicitly demonstrates that this state can be realized on the surface of a 3d bulk system while retaining @xmath0 and charge u(1 ) symmetries .
since the surface topological order is forbidden in a strictly 2d system , we have realized _ a _
3d topological insulator .
moreover , this phase has magnetoelectric response @xmath3 .
the remaining question is - is this the same phase as the free fermion topological insulator ? a necessary condition to make
this identification is that in the absence of charge conservation symmetry ( i.e. induced by proximity coupling the surface to a superconductor ) , one should recover the ti surface superconductor , without topological order , but with majorana zero modes in the vortex cores .
the t - pfaffian state however allows no simple way to exit the topological phase even when charge conservation is absent , while retaining @xmath0 . on the other hand
, we find that there are two versions of the t - pfaffian state ( t - pfaffian@xmath20 , with @xmath21 ) which differ in the way the nonabelian particles transform under time reversal symmetry .
the non - abelian particles with bosonic ( fermionic ) topological spin is assigned @xmath22 ( @xmath23 ) .
we can then demonstrate the following fact : ( i ) the two states corresponding to @xmath24 and @xmath25 differ by the surface topological order of a bosonic topological superconductor ( btsc ) with a @xmath26 classification and ( ii ) they can differ from the 3d free fermion ti surface at most , by the same btsc surface topological order .
this implies that one of them _ must be _ the free fermion ti surface state while the other represents a mixture with a btsc although , unfortunately , we can not at present specify which of the @xmath21 is the 3d ti surface .
we also discuss a second topological order , the pfaffian - antisemion state obtained recently @xcite by a series of elegant physical arguments .
this state is a tensor product of the read - moore pfaffian state as discussed above , with a neutral anti - semion theory @xmath27 and has 24 quasiparticles .
its statistics is compatible with @xmath0 symmetry , and furthermore passes the necessary requirements for being identified with the 3d ti surface state , including realizing the ti surface superconductor on breaking charge conservation @xcite .
here we prove that it is indeed realizable as the surface state of a 3d bulk system with the requisite symmetries , by constructing an exactly soluble 3d lattice model . in both this and the t - pfaffian case , we find time reversal symmetry is respected _ only _ if the electrons transform projectively , i.e. as @xmath4
. a central tool will be the walker - wang construction@xcite of an exactly soluble lattice hamiltonian , that realizes the desired surface topological order , while maintaining a topologically trivial bulk .
the model realizes a ground state wave function which is a superposition of 3d loops , one for each quasiparticle of the surface topological phase .
the amplitude for any configuration is obtained as follows .
the topological order is represented in terms of the @xmath28 and @xmath29 symbols that are associated with certain basic loop moves@xcite .
first , one projects the loop configuration onto a 2d plane , and relates it to a reference configuration using the elementary moves . for each move ,
the amplitude acquires a factor that is related to the @xmath28 and @xmath29 symbols .
based on the resulting state one can readily show that an anyon that has nontrivial mutual statistics with some other excitation , is necessarily confined to the surface .
furthermore , the surface excitations realize the required topological order . a minor caveat here is that for simplicity we work with bosonic walker wang models , without elementary fermions in the hilbert space@xcite . since the surface topological order is non - modular , i.e. it contains a fermion excitation that has trivial mutual statistics with everything else , this excitation is deconfined in the bulk .
hence the 3d state that is realized is a z@xmath30 gauged version of the topological insulators , i.e. it involves bulk fermions that carry gauge charge and loops carrying @xmath31 gauge flux
. this state may thought of as being obtained from a bosonic model , from a parton construction @xmath32 , where the z@xmath30 gauge charged fermions @xmath33 are placed in a 3d topological phase@xcite .
this can be readily rectified by introducing elementary fermions @xmath34 and condensing the product @xmath35 which confines the gauge flux and removes the bulk topological order . before moving to the technical details
we raise the following question that may have occurred to some readers . how is the @xmath3 magneto electric response of 3d topological insulator reconciled with the topologically ordered insulating surfaces with @xmath0 symmetry ?
note , @xmath3 implies that a weak applied magnetic field produces a surface charge density of @xmath36 ( modulo integer multiples of @xmath17 ) per flux quantum .
of course , such a charge density is meaningful only if the surface is also insulating . also , charge is only determined modulo @xmath17 since integer multiples of @xmath17 can be screened by surface electrons .
the fractional part however can not be screened by electrons and is a bulk property . specifying whether this charge density is @xmath37 leads to the usual argument for breaking of time reversal symmetry at the surface .
the key observation here is that the candidate states both contain charge @xmath36 excitations which can screen the induced charge , and hence breaking of @xmath0 symmetry is not required .
first let us introduce the topological order in the t - pfaffian state , including the anyons types and their fusion and braiding rules .
the t - pfaffian state is a twisted version of the pfaffian state ( such that the state could potentially be time reversal invariant ) .
similar to pfaffian , it is the combination of the non - abelian ising theory with the abelian @xmath38 theory .
the ising theory describes a gauged a @xmath39 superconductor with @xmath11 fluxes and contains anyons labeled by @xmath40,@xmath41,@xmath42 .
@xmath40 labels the trivial vacuum , @xmath42 is the fermion in the superconductor and @xmath41 is the @xmath11 flux .
they fuse according to @xmath43 @xmath41 is hence nonabelian and has quantum dimension @xmath44 while @xmath42 has quantum dimension @xmath45 .
the topological spins for the anyons are _ i=1 , _
= e^-i , _ = -1 the braiding of the fermion @xmath42 around the @xmath11 flux results in a phase factor of @xmath46 .
the ising sector is neutral and does not carry charge .
the @xmath38 theory describes the @xmath47 quantum hall state where charge @xmath48 electron pairs form an effective @xmath49 bosonic laughlin state .
the chern - simons effective theory for this state is = a__a_^ there are @xmath50 abelian anyons in the theory labeled by @xmath51 which add in the usual way when they fuse k_1k_2= ( k_1+k_2 ) 8 the topological spins for the anyons are given by _
k = e^ik^2 a full braiding of particle @xmath52 around particle @xmath53 results in a phase factor of _ kk=e^ikk the @xmath38 sector carries fractional charge with the @xmath52 particle carrying @xmath54 charge .
the t - pfaffian theory is then obtained by combining @xmath40 and @xmath42 of the ising theory with even @xmath52 of the @xmath38 theory and @xmath41 with odd @xmath52 .
that is , the t - pfaffian theory has anyons llll i_0 & i_2 & i_4 & i_6 + _ 0 & _ 2 & _ 4 & _ 6 + _ 1 & _ 3 & _ 5 & _ 7 the charge assignment of the t - pfaffian theory is carried over from the @xmath38 theory .
the fusion and braiding statistics of the combined anyons is the product of the fusion and braiding of the ising part and the @xmath1 part . in particular ,
the topological spin for the combined anyons are i & 1 & & i & & 1 & & i & + & & 1 & & -1 & & -1 & & 1 + & -1 & & -i & & -1 & & -i & + & & & & & & & & + charge & 0 & e/4 & e/2 & 3e/4 & e & 5e/4 & 3e/2 & 7e/4 + [ t_spin ] the quantum dimensions of the @xmath55 and @xmath56 ( even @xmath52 ) anyons are @xmath45 and that of the @xmath57 ( odd @xmath52 ) anyons are @xmath44
. the particle and anti - particle pairs in the t - pfaffian theory are i_k ~i_(8-k ) 8 , _ k ~_(8-k ) 8 , _ k ~_(8-k ) 8 obviously @xmath58 is the vacuum in the theory .
moreover , it is easy to check that a full braiding of @xmath59 around any other particle in the theory leads to a phase factor of @xmath45 .
that is , the @xmath59 particle is a local excitation of the system and can not be seen by a braiding operation far away .
in fact , @xmath59 has topological spin of @xmath46 , carries charge @xmath17 and is therefore the electron in the system .
note that while the t - pfaffian state has the same anyon types as the pfaffian state , the statistics of the two are different .
in particular , the the statistics of the ising part is taken to be the complex conjugate of that in pfaffian .
therefore , for example , the topological spin of the @xmath60 anyon is @xmath45 in the t - pfaffian theory while it is @xmath61 in the pfaffian theory .
is it possible to realize the t - pfaffian topological order in a time reversal invariant system ? for pure 2d system , the answer is no .
this is easy to see if we look at the edge of the system .
the ising part of the theory has chiral central charge @xmath62 and is neutral , hence does not contribute to charge hall conductance @xmath63 . on the other hand ,
the @xmath38 part of the theory has chiral central charge @xmath64 and @xmath65 .
therefore , the t - pfaffian theory has total chiral central charge @xmath66 and charge hall conductance @xmath65 .
obviously this is not possible in a pure 2d system with time reversal symmetry .
however , such argument fails if the t - pfaffian theory is realized on the surface of a 3d gapped system , because the 2d surface of a 3d system does not have an edge of its own .
therefore , the chiral edge of the t - pfaffian state no longer presents an obstruction to realization in a time reversal invariant system . on the surface of a 3d system
, we can only probe how time reversal symmetry acts on the excitations in the bulk of the 2d system , namely the anyons .
first , the action of time reversal symmetry takes complex conjugation of all the braiding and fusion processes among the anyons .
moreover , time reversal symmetry can map one anyon type to another .
therefore , if the t - pfaffian theory can be realized on the surface of a 3d time reversal invariant system , then the time reversal symmetry must exchange the anyon types in such a way that the fusion and braiding amplitudes are invariant under both complex conjugation and this exchange of anyon types .
such an exchange of anyon type does seem to exist if we consider the topological spins of the anyons , which describe the self rotating processes of the anyons . from eqn.[t_spin ] , we can see that the topological spins remain invariant if the time reversal symmetry performs the following exchange i_2 _ 2 , i_6 _ 6 together with taking complex conjugation .
compare to the pfaffian state where such an exchange of anyon type , hence time reversal symmetry , can not exist .
in particular , the topological spins for the four @xmath41 anyons in the pfaffian theory are _ _ 1=e^i , _ _ 3=e^i , _ _ 5=e^i , _ _ 7=e^i , which do not form time reversal invariant pairs .
moreover , we can check that the exchanges @xmath67 with @xmath68 and @xmath69 with @xmath70 is consistent with the fusion rules of the t - pfaffian theory .
for example , the fusion process of i_2 _ 4 = _ 6 is mapped to _ 2 _ 4 = i_6 under this exchange , which is again a valid fusion process in the t - pfaffian theory .
therefore , we have found an exchange of anyon types which , together with complex conjugation , keeps the fusion rules and the topological spins of the t - pfaffian theory invariant .
the exchange of anyon types , however , does not completely specify the action of time reversal symmetry on the t - pfaffian state .
in particular , for anyon types which do not change under time reversal , one can ask whether time reversal acts as @xmath71 or @xmath46 on the anyon locally . this is a legitimate question to ask , because away from the anyonic excitations , the state remains invariant under time reversal .
if anyon types does not change under time reversal , then time reversal acts effectively locally around the anyon . as shown in
, the local action of time reversal can only square to @xmath45 or @xmath46 .
if @xmath72 on a particular anyon , then then anyon has an extra spin label and leads to a local @xmath73 fold degeneracy under time reversal symmetry when it is separated from all other anyons . on the other hand ,
if the anyon changes type under time reversal , then the effective action of time reversal is nonlocal on the state and it is not well defined to talk about @xmath74 locally for the anyon .
the local time reversal symmetry action represented by this @xmath75 information is important because with different local action , the topological state can correspond to totally different bulk phase .
for example , the statistics of @xmath11 gauge theory with anyons @xmath76 is time reversal invariant with no exchange of anyons . if the two bosonic particles @xmath17 and @xmath77 both transform as @xmath71 , the state can be realized in 2d time reversal invariant system . however , if they both transform as @xmath72 , the state can only be realized on the surface of 3d bosonic topological superconductors@xcite . for the t - pfaffian state , we would be interested in the @xmath74 transformation for the charged boson @xmath78 , the chargeless fermion @xmath79 , the electron @xmath59 and all the nonabelian @xmath57 particles .
in particular , the @xmath74 transformation of @xmath59 would tell us whether we are dealing with a @xmath71 or @xmath46 topological insulator . in this section , we state the general rules for determine the @xmath74 information for anyons .
we give the motivation for setting these rules and apply them to the t - pfaffian state . in appendix [ fr_t ] , we will provide an algebraic proof for these rules in terms of the exactly solvable walker - wang model realizing the particular topological state on the surface .
the local action of time reversal on the anyons can be determined from the following two rules that applies to both abelian and non - abelian anyons : * * rule 1 : * if anyons @xmath80 and @xmath81 fuse into @xmath52 and none of @xmath80 , @xmath81 , @xmath52 change type under time reversal , then @xmath82 . * * rule 2 : * if @xmath80 maps into @xmath83 ( different from @xmath80 ) under time reversal and the braiding of @xmath80 around @xmath83 in the fusion channel @xmath52 resulting in a phase factor of @xmath84 , then @xmath85 . the first rule comes from considering a region with two anyons @xmath80 and @xmath81 .
if @xmath80 and @xmath81 are separated far enough ( larger than correlation length ) , then they each have well defined @xmath74 .
the total local time reversal symmetry action on the whole region is composed of that on @xmath80 and @xmath81 separately .
however , for an observer very far away from this region , the total anyonic charge in the region is @xmath52 .
therefore , @xmath86 .
note that one consequence of this rule is that a particle and its antiparticle have the same @xmath74 .
the second rule comes from considering a region with @xmath80 and @xmath83 . applying time reversal , @xmath80 changes into @xmath83 and vice versa .
this is equivalent to rotating the @xmath87 pair by @xmath88 degree . applying time reversal twice
, we have rotated the pair by @xmath89 degree which is equivalent to a full braiding of @xmath80 around @xmath83 and results in a phase factor of @xmath84 . however , for an observer far away from this region , the total anyonic charge in the region is @xmath52 and the @xmath84 phase factor comes from @xmath90 . applying these two rules
, we can find out possible @xmath74 action on each anyon locally .
there may be more than one set of @xmath74 values for all the anyons satisfying these rules .
let s now apply these rules to t - pfaffian and determine the local time reversal symmetry action .
first of all , the electron @xmath59 is the fusion product of @xmath67 and @xmath68 which map into each other under time reversal . because the braiding of @xmath67 around @xmath68 gives a @xmath46 , @xmath59
transform as @xmath72 .
therefore , we are indeed dealing with a @xmath72 topological insulator .
next , because @xmath60 and @xmath91 can fuse both to @xmath59 and @xmath78 and they do not change type under time reversal , using the first rule we find that @xmath78 transforms in the same way as @xmath59 .
that is , @xmath72 for the charged boson .
from this , we can easily tell that @xmath79 , the chargeless fermion , transforms as @xmath71 .
finally , @xmath92 have the same @xmath93 , while @xmath94 have @xmath95 .
this information is summarized in the following table of @xmath74 values : i & 1 & & & & -1 & & & + & & & & - & & - & & + & 1 & & & & -1 & & & + [ t_tsquared ] _ two t - pfaffians : _ hence there are two possible ways that time reversal symmetry can act on the t - pfaffian state , labeled with @xmath96 .
these two states can be mapped into each other by combining with the following @xmath11 gauge theory .
consider a @xmath11 gauge theory where the gauge charge @xmath97 and the gauge flux @xmath98 both transform as @xmath72 and @xmath97 carries @xmath1 charge @xmath46 ( in units of the electron charge ) while @xmath98 is neutral .
this particular @xmath11 gauge theory can not be realized in a purely 2d system as discussed in ref.@xcite , where it was termed the etmt state , and realizes the surface topological order of a bosonic spt phase protected by time reversal symmetry ( charge attached to the @xmath97 particle is an unimportant difference ) .
bring such a @xmath11 gauge theory on top of the @xmath99 t - pfaffian state and condense the boson pair @xmath100 .
the combination of @xmath97 and @xmath78 is charge neutral and transforms as @xmath71 , therefore , the condensate preserves both symmetries . after the condensation , the gauge flux @xmath98 gets bounded to the @xmath101 particles in the t - pfaffian state in order to commute with the condensate while all the abelian particles in the t - pfaffian state remains .
therefore , the particle content in the resulting theory is the same as the original t - pfaffian state , but the time reversal representation of the @xmath101 particles change from @xmath99 to @xmath25 .
the @xmath1 charge carried by the particles remains the same .
thus , these two states differ by a particular bosonic topological superconductor .
from the discussion in the previous sections we see that a possible definition of time reversal and charge conservation symmetry action does exist for the anyonic excitations in the t - pfaffian state .
therefore , the t - pfaffian state could potentially be realized on the surface of 3d systems with@xmath0 and charge conservation symmetry , although not in purely 2d due to the chiral edge modes in t - pfaffian . in the next section ,
we show that such a 3d realization indeed exists by presenting an exactly solvable model using the walker - wang construction@xcite . due to the nontrivial symmetry action in t - pfaffian
, the 3d bulk of the system must have some nontrivial symmetry protected topological order .
that is , the model is a 3d topological insulator .
but is it _ the _ topological insulator realized in free fermion systems or some other previously unknown topological insulator which is only possible in strongly interacting systems ? to answer this question definitely , we would need to find some topological invariants for different topological insulators and compute them for this system .
however , we do not know how to do this . in the following
, we will check certain properties of this model and see if it is consistent with what we know about the free fermion topological insulator .
we find that : 1 .
two copies of t - pfaffian is trivial which is consistent with the @xmath11 classification of the free fermion topological insulator ; 2 . by breaking time reversal but not charge conservation symmetry , the topological order can be removed .
now , surface domain walls between regions with opposite @xmath0 breaking carry gapless 1d modes with @xmath64 and @xmath102 , which is known to happen in the free fermion topological insulator .
this also implies @xmath3 ; 3 . for one of the t - pfaffian states ( @xmath99 or @xmath46 ) , the topological order can be removed by breaking charge conservation but not time reversal symmetry , a property expected for free fermion topological insulator surface states .
while we do not explicitly construct the route to removing topological order , we demonstrate this to be a logical consequence .
the free fermion topological insulator ( ti ) has a @xmath11 classification .
that is , if we take two copies of the free fermion ti and allow interactions between them the surface state can be made trivial without breaking either time reversal or charge conservation symmetry .
therefore , if the t - pfaffian state is realized on the surface of the free fermion ti , we should be able to take two copies of it and removed the topological order without breaking time reversal or charge conservation symmetry .
this is indeed the case as we show below .
suppose that we have two t - pfaffian states whose anyons are labeled as @xmath103 and @xmath104 .
we can condense the following set of composite bosonic particles without breaking time reversal or charge conservation : i_2_6,_2_6,i_6_2,_6_2,i_4_4,_0_0,_4_4 note first that each composite particle listed above has bosonic self and mutual statistics , therefore they can be condensed together .
also , each composite particle has charge @xmath105 , hence condensing them does not break charge conservation .
moreover , the composite particles either map to themselves under time reversal or appear in time reversal pairs . finally , they all transform as @xmath71 under time reversal .
therefore , the condensate does not break time reversal either .
after condensing these particles , the nonabelian @xmath57 and @xmath106 particles are all confined .
some of the composite @xmath107 type particles remains , which up to the condensed particles include _ 1_3 , _
1_7 the abelian particles that remain include ( up to the condensed particles ) i_4,_0,_4 in the resulting theory , the @xmath107 particle splits into two abelian particles and the theory is equivalent to the product of a free fermion part @xmath108 and a simple @xmath11 gauge theory part i , e , m,@xmath17 and @xmath77 comes from @xmath109 .
they are bosons , carry charge @xmath110 and are invariant and transform as @xmath71 under time reversal .
@xmath111 comes from @xmath79 .
it is a fermion , has charge @xmath110 and maps to itself and transform as @xmath71 under time reversal .
therefore , the total theory is trivial under time reversal and charge conservation symmetry and can be realized in 2d . to remove the topological order in the t - pfaffian surface state by breaking time reversal but not charge conservation symmetry ,
we can bring a 2d fractional quantum hall state with the t - pfaffian topological order and couple it to the t - pfaffian surface state .
the 2d state has @xmath112 and @xmath2 , therefore is not time reversal invariant .
but it does have charge conservation symmetry .
we label the anyons in the surface t - pfaffian state as @xmath103 and those in the 2d t - pfaffian state as @xmath113 .
condense the following composite particles : i_2_6,_2i_6,i_6_2,_6i_2,i_4i_4,_0_0,_4_4 note that this condensation is very similar to the one we applied in the previous section to two copies of t - pfaffian surface states . however , here the operation breaks time reversal from the beginning because the 2d t - pfaffian state breaks time reversal symmetry . after this condensation ,
the surface state is reduce to the product of a charge neutral @xmath11 gauge theory with anyons @xmath76 together with a charged fermion .
we can further remove the topological order by condensing the @xmath17 particle in the @xmath11 gauge theory . to break the time reversal symmetry in the opposite way and remove the topological order , we bring a time reversed copy of the 2d t - pfaffian state with anyons @xmath114 and couple it in the time reversed way to the surface t - pfaffian state .
the statistics in the time reversed copy of the 2d t - pfaffian is the complex conjugate of that in the original 2d t - pfaffian state .
therefore , in this new combination , we would condense _
2|_6 , i_2|i_6,_6|_2,i_6|i_2,i_4|i_4,_0|_0,_4|_4 the resulting theory is again composed of a neutral @xmath11 gauge theory with @xmath115 and a charged fermion . by condensing @xmath116 ,
we remove the topological order completely . between the 2d t - pfaffian state and its time reversal copy ,
there is a @xmath64 and @xmath117 edge .
condensation in the system does not affect @xmath118 and @xmath63 .
therefore , we can break time reversal symmetry in opposite ways on the surface t - pfaffian state , remove any topological order , and be left with a @xmath64 and @xmath117 chiral edge between the two regions .
this is what is known to happen on the free fermion ti surface starting from the gapless dirac cone surface state .
when the surface of the free fermion topological insulator is gapless , the surface dirac cone can be gapped out ( in a topologically trivial way ) by inducing superconductivity on the surface without breaking time reversal symmetry .
if the t - pfaffian can be realized as the topologically ordered surface state of the free fermion ti , we would like to see that the topological order can be removed by condensing charge without breaking time reversal symmetry . in the t - pfaffian state
, it is not obvious how this can be achieved .
for example , one might want to condense the charged boson @xmath78 and simplify the topological order .
however , @xmath78 transforms under time reversal as @xmath72 .
therefore , condensing @xmath78 necessarily breaks time reversal symmetry .
the other abelian particles are not bosons and can not be directly condensed .
we show that such a removal of topological order can be achieved for one of the t - pfaffian state ( @xmath99 or @xmath25 ) by combining with a 2d topological order which has time reversal but not charge conservation symmetry and then condensing composite bosonic particles . therefore , one of the t - pfaffian state is consistent with being the surface state of the free fermion ti .
our argument proceeds in the following steps : .
the ` modularized ' t - pfaffian is a time reversal symmetric bosonic topological state . as a fermionic topological order
, the t - pfaffian state has @xmath11 fermion parity symmetry .
we can gauge the @xmath11 symmetry and obtain a ` modularized ' bosonic topological theory where the local fermion has a mutual @xmath46 statistics with the @xmath11 fluxes .
such a gauging process is not unique and we find that one of the possible ` modularized ' theories has time reversal symmetric fusion and braiding statistics . the gauging process and
the resulting theory is described in detail in appendix [ gauge ] .
this step works for both versions of the t - pfaffian ( @xmath96 ) .
one of the ` modularized ' t - pfaffians can be realized in 2d with time reversal symmetry . as a bosonic topological order with time reversal symmetric fusion and braiding statistics ,
the ` modularized ' t - pfaffian states can be either realized in 2d time reversal symmetric systems or on the surface of 3d bosonic topological superconductors . by simply looking at the theories , it is hard to tell which is the case .
however , useful information can be obtained from our knowledge of bosonic topological superconductors .
we know that bosonic topological superconductors has a @xmath119 classification@xcite .
these are composed of ( i ) the nontrivial state in the first @xmath11 has three fermion surface topological order ( which is chiral when realized in 2d ) while ( ii ) the nontrivial state for the second @xmath11 has both the electric and magnetic charges transforming as @xmath4 and is a nonchiral surface topological order@xcite , labeled etmt in ref.@xcite .
because the modularized t - pfaffian theories are nonchiral , they must belong to either the trivial or the nontrivial case of the second @xmath11 ( i.e. etmt surface topological order ) .
moreover , the two modularized t - pfaffians differ by exactly etmt . to see this ,
take the etmt surface topological order and bring it on top of the @xmath99 modularized t - pfaffian .
condense the composite particle @xmath120 .
because both @xmath17 and @xmath78 are both kramer doublets , such a condensation does not violate time reversal symmetry . the resulting theory after the condensation is exactly the @xmath25 modularized t - pfaffian state . therefore , one of the modularized theory can be realized in 2d with time reversal symmetry while the other is the surface state of a nontrivial bosonic topological superconductor ( etmt surface topological order ) .
one last piece of missing information is which one is which .
we do not know the answer to this question right now .
\3 . combining the ` modularized ' t - pfaffian with t - pfaffian
, the topological order can be removed without breaking time reversal symmetry .
now bring the modularized t - pfaffian state ( complex conjugated ) on top of the corresponding t - pfaffian state .
condense fermion - fermion pair to confine the @xmath11 gauge field and we obtain a doubled t - pfaffian theory . the topological order in this doubled theory
can be completely removed by condensing pairs of corresponding particles , e.g. @xmath121 , which does not violate time reversal symmetry .
therefore , for one of the t - pfaffians , the surface topological order can be removed through combination with a 2d @xmath122 invariant topological state and condensing bosons in a time reversal invariant way .
therefore , we can conclude from the previous analysis that one of the t - pfaffian states , on breaking charge conservation symmetry , is equivalent to the superconducting surface of the free fermion topological insulator , while the other differs from it by a bosonic topological superconductor ( with surface topological order etmt ) .
thus one of the t - pfaffian states passes all the physical requirements expected of ti surface topological order - z@xmath30`ness ' , , thermal and electrical hall ( equivalent to @xmath3 ) conductivity on @xmath0 breaking surface domain walls , and surface superconductor with @xmath0 symmetry and free of topological order .
however at this moment , we can not tell whether this is the @xmath99 state or the @xmath25 t - pfaffian state .
we mention here a simple argument that allows us to connect one of the t - pfaffian states with the free fermion ti , based on the classification result of .
we note however that the argument above did not utilize this result . in ref . , @xmath123 fermionic topological insulators in 3d
were proposed to have a @xmath124 classification - one @xmath11 corresponds to the free fermion topological insulator while the remaining @xmath125 refers to neutral bosonic spts with time reversal symmetry .
the ambiguity here is whether the t - pfaffian(s ) describe a mixture of the free fermionic topological insulator with 3d bosonic spt phases .
we can prove that there is a t - pfaffian state which is to be identified with the free fermion ti , with no bosonic spt mixture assuming the classification result above .
first let us discuss possible bosonic spt phases with @xmath126 symmetry which have a @xmath124 classification .
the root states in terms of surface topological order are ( i ) three fermion state ( ii ) the toric code with @xmath17 and @xmath77 transforming as @xmath72 ( etmt state ) and ( iii ) the toric code where both @xmath17 and @xmath77 are charge @xmath127 of the cooper pair . however ,
in the presence of electrons the classification is reduced to @xmath125 for these bosonic spts .
one can combine say the electron with ( iii ) to obtain a mixture of ( i ) and ( ii ) .
therefore one can take the two neutral states ( i ) and ( ii ) as the relevant topological orders @xcite .
the remaining question is - is the t - pfaffian a mixture of fermionic ti and one or both of the bosonic spts ( i ) , ( ii ) ?
note , if it is a mixture , then breaking charge conservation symmetry is not sufficient to eliminate the topological order .
the t - pfaffian can not contain the state ( i ) since t - pfaffian has @xmath128 and @xmath129 in one realization while the mixture with state ( i ) would have an additional thermal hall conductivity of @xmath130 , while retaining the same charge hall conductivity . here
@xmath131 and @xmath132 where @xmath133 is the lorentz number and @xmath134 is the temperature . therefore the t - pfaffian can at best differ from the free fermion ti by the etmt state ( ii ) .
however , the two versions of the t - pfaffian differ from one another by exactly this state , which has a @xmath11 classification .
hence one of the two t - pfaffian states represents the surface of the free fermion ti .
in this section , we construct the 3d model with time reversal symmetry which realizes the t - pfaffian theory on its surface and have time reversal acting in the way expected . in section [ wwa ] , we introduce the basic idea of the walker wang construction and explain in section [ wwt ] how it allows us to determine the local time reversal symmetry transformation ( @xmath75 ) for the anyons .
we try to first present the basic picture and the general idea underlying the walker wang construction in this section without going into too much details , which is saved for section [ wwd ] where we give the exactly solvable hamiltonian and address some related subtleties .
the walker - wang construction provides a way to write down an exactly solvable 3d model which realizes a particular topological order on the 2d surface of the system@xcite .
given all the fusion and braiding data of a 2d anyon theory , the walker - wang prescription gives the local hilbert space , terms in the hamiltonian and ground state wave function of a 3d model such that the 2d anyon theory emerges on the surface of the system . while it is not surprising that 2d anyon theories can be realized on the surface of 3d systems , the walker - wang construction is useful in the following ways : ( 1 ) it provides exactly solvable 3d models to realize ` chiral ' 2d topological orders , for which a 2d exactly solvable model is not known to exist ; ( 2 ) the 3d walker - wang model can have extra symmetry than is possible on the topological order in a purely 2d system .
that is , the surface of the walker - wang model can realize symmetry enriched topological orders that is not possible in 2d , which is a result of the nontrivial symmetry protected topological order in the 3d bulk of the system . in our previous works , we have explored this property of the walker - wang model in the case of bosonic and fermionic topological superconductors , demonstrating the existence of time reversal invariant topological orders which are impossible in purely 2d systems but can be realized on 3d surface . here , we use a similar strategy to study fermionic topological insulators and show that the t - pfaffian state can be realized on the surface of a 3d system with time reversal and charge conservation symmetry while it is not possible in 2d with the same symmetry . in this section
, we are not going to explain all the details related to the exactly solvable hamiltonian , but only focus on the basic idea of the walker - wang construction and show how it allows us to determine that @xmath72 for the electrons .
( @xmath74 for other anyons can also be determined . )
the basic idea underlying the walker - wang construction is very intuitive .
the model is constructed such that the ground state wave function is a superposition of 3d loops ( more precisely ` string nets ' , in the sense of @xcite ) labeled by the anyon types , which describes the 2 + 1d space time trajectory of the anyons .
the amplitude for a given configuration of these loops @xmath135 in the 3 + 1d wave function @xmath136 is determined by the expectation value of the corresponding wilson loop operators in the 2 + 1d tqft ( topological quantum field theory ) which describes how the anyon world lines twist and intertwine with each other ; i.e. we have : _
3d(c ) = w(c)_2 + 1tqft [ ground_state ] this is similar in spirit to eg . quantum hall wave functions , which are related to the space - time correlations of their edge states . here , since we demand a topologically ordered boundary state , the expectation values are taken in the boundary tqft .
the wave function for the t - pfaffian walker - wang model hence contains 12 different string types corresponding to the 12 different anyons in the theory which can braid and fuse according to the fusion rules of t - pfaffian .
the strings have directions .
if the direction of a string related to anyon @xmath80 is reversed , it becomes a string related to the anti - particle , @xmath137 .
since the twisting and intertwining of the anyon world lines may depend on the angle of view , in order to calculate the amplitude of the string - net configurations , we need to pick a particular projection of the 3d loops onto a 2d surface .
the projection we will use is also shown in fig.[trivalent_lattice ] .
having fixed a projection , the amplitude of each configuration can be obtained using the braiding and fusion rules given by the @xmath28 and @xmath29 matrix of the t - pfaffian theory , which is the product of the @xmath28 and @xmath29 matrices of the ising part and the @xmath38 part .
the @xmath28 matrix for the ising part reads l r^i,*_*=r^*,i_*=1 , r^,_i = e^i , r^,_=e^-i + r^,_=r^,_=i , r^,_i=-1 the @xmath28 matrix in the @xmath1 part is r^k_1,k_2_(k_1+k_2 ) 8=e^ik_1k_2 the @xmath29 matrix for the ising part is l _ , = 1 & 1 + 1 & -1 + _ , = _ , = -1 where @xmath138 .
all other terms being @xmath45 .
the @xmath29 matrix for the @xmath1 part is l _ ( k_1+k_2 )
8,(k_2+k_3 ) 8 + = e^ik_1(k_2+k_3-(k_2+k_3 ) 8) it can only take value @xmath139 .
using the braiding and fusion moves as illustrated in fig.[rf ] , we can deform any string - net configurations configuration to a set of isolated loops .
the change in the amplitude of the configurations is given by the @xmath29 and @xmath28 matrices .
the isolated loops can be removed with the change in amplitude by l _ i_k=__k=1 , k=0,2,4,6 + _ _ k=- , k=1,3,5,7 using these set of rules , the amplitude of any string - net configuration can be determined ( relative to the all @xmath58 configuration ) .
such a bulk wave function encodes the statistics on the surface , as we show below .
anyonic excitations can be created by adding open strings to the surface .
the wave function becomes a superposition of all string - net configurations in which the corresponding strings end at the positions of the excitations .
then we can check the statistics of the excitations by tracking these open strings .
suppose we exchange two string ends of the same type @xmath140 ( as show in fig.[surface_anyon ] ( a ) ) by crossing two red string segments on the surface .
the two @xmath140 anyons fuse to a @xmath141 anyon .
( fig.[surface_anyon](a ) shows one possible string - net configuration . )
this twist in the string - net configuration ( relative to the string - net configuration before exchange ) can be removed to bring the strings back to their original form , but this results in a factor of @xmath142 .
therefore , exchanging end of strings of the same type adds a @xmath142 factor to the total wave function , which is equivalent to saying that the ends of the strings are anyons with self statistics given by @xmath142 .
similarly one can check , with linked loops on the surface as shown in fig.[surface_anyon ] , that string ends of different types have mutual statistics given by the corresponding @xmath28 matrix element .
are anyonic excitations of type @xmath140 with the expected statistics .
this can be seen from the braiding statistics of the strings generating ( a ) the exchange and ( b ) the braiding of the end of strings . ]
open strings in the bulk also create excitations in the ground state . however ,
if the corresponding anyon has nontrivial braiding with any other anyon , the excitation energy grows linearly with the string length , leading to confinement of the particles at the ends of the strings . to see the confinement , consider an open string of type @xmath140 ( colored blue ) in the bulk which is circled by a small ring of a different string type @xmath141 ( colored red ) , as shown in fig.[confinement ] has nontrivial braiding with some other anyon @xmath141 , then the open string of type @xmath140 in the bulk can change the quantum fluctuation phase factors of @xmath141 loops along its length , which costs finite energy . therefore ,
the end of strings of type @xmath140 are confined in the bulk .
otherwise , the end of string of type @xmath140 is a deconfined point excitation , which must be either bosonic or fermionic in a 3d system . ] using the braiding rule between the loops we find that the linking between the ring and the open string can be removed together with a phase factor s_,= _ d _ r^|_r^| _ where @xmath143 is the fusion product of @xmath140 and @xmath141 , @xmath144 is its quantum dimension and @xmath145 . if @xmath146 , the open string changes the quantum fluctuation phase factors of small loops along its length , which costs finite energy .
therefore , the string s endpoints can not be separated very far , and the corresponding anyonic excitations in the bulk are confined . if @xmath147 for all @xmath141 , then the end of string of type @xmath140 can be a deconfined point excitation in the 3d bulk of the system , which has to be either a boson or a fermion .
in the t - pfaffian state , all anyons have nontrivial braiding with some other anyon except the electron @xmath78 .
the electron is a local excitation in the 2d topological order and hence has trivial braiding with any other anyon .
therefore , in the t - pfaffian walker - wang model , the only deconfined excitation in the 3d bulk is the electron .
however , open strings lying on the surface , where no loops can encircle them , give rise to deconfined excitations@xcite .
therefore , the 3d walker - wang model written in terms of the fusion and braiding rules of the t - pfaffian theory has a deconfined electron in the bulk and deconfined quasiparticles corresponding to all the anyons in the t - pfaffian theory on the surface .
we will postpone describing all details of the exactly solvable model to section [ wwd ] .
first , let me see how the walker - wang construction tells us more about time reversal symmetry action in the t - pfaffian state .
the walker - wang model provides us with not only an exactly solvable model to realized the t - pfaffian surface state , but also a more concrete setup to study the time reversal symmetry in the system . because the ground state wave function is determined by the @xmath29 and @xmath28 matrices of the t - pfaffian state , in order for the wave function to be time reversal symmetric
, the time reversal symmetry action must leave the @xmath29 and @xmath28 matrices invariant .
however , a quick check shows that the @xmath29 and @xmath28 matrices are not invariant under the exchange of @xmath148 , @xmath149 and complex conjugation .
for example , and @xmath28 such that simple exchange and complex conjugation leaves them invariant . ]
( r^_2i_2__4)^*=-r^i_2_2__4 in order to fix this , we need to introduce extra phase factors @xmath150 in the time reversal symmetry action to the vertices where three strings meet . with proper choice of @xmath150 ,
the @xmath29 and @xmath28 matrices can be invariant as l ^*_m , n=_|m,|n + ( r^ij_k)^*=r^|i|j_|k [ fr_t_eq ] where @xmath83 is the time reversal partner of @xmath80 .
that is , we need to introduce some extra vertex degrees of freedom at the branching point of strings , which gets a phase factor under time reversal symmetry action . with a proper choice of gauge for @xmath29 and @xmath28 ( explained in detail in appendix [ fr_t ] ) , a possible set of @xmath151 s is : @xmath152 for the vertices l ( _ 1,_5,i_2 ) , ( _ 3,_7,i_6 ) , ( i_2,i_2,i_4 ) , + ( i_2,_2,_4 ) , ( i_6,i_6,i_4 ) , ( i_6,_6,_4 ) [ t1 ] and @xmath153 at their time reversal partners l ( _ 1,_5,_2 ) , ( _ 3,_7,_6 ) , ( _ 2,_2,i_4 ) , + ( _ 2,i_2,_4 ) , ( _ 6,_6,i_4 ) , ( _ 6,i_6,_4 ) [ t2 ] and @xmath154 for all other allowed vertices . note that here we are labeling the strings at each vertex such that the corresponding anyons fuse to the vacuum , i.e. the direction of the strings are all pointing towards the vertex .
the strings at each vertex are ordered in a clockwise way .
( with this choice of @xmath151 , eq . [ fr_t_eq ] is satisfied , not for the @xmath29 and @xmath28 given above but with some other gauge choice of @xmath29 and @xmath28 .
this is explained in detail in appendix [ fr_t ] . ) with this choice of @xmath151 , we find that @xmath72 on all the vertices listed above in eq .
[ t1 ] and [ t2 ] , while @xmath71 on all other allowed vertices .
but where are these @xmath72 vertex degrees of freedom coming from ?
actually they are related to the @xmath75 transformation law for each anyon type under time reversal , as we explain below .
an simple way to understand the vertex @xmath72 degrees of freedom is to ` attach haldane chains ' to the strings labeled by @xmath91 , @xmath155 , @xmath78 and @xmath59 .
imagine adding pairs of spins @xmath127 to each string segment .
along the strings of types _ 3 , _ 5 , i_4 , _ 4 the spin @xmath127 s are put into a ` haldane chain ' state where spins on neighboring string segments are connected into singlet pairs , as shown in fig.[hchain ] . along strings of all other types ,
the two spins on the same segment are connected into singlet pairs .
a sample configuration is shown in fig .
[ hchain ] .
s , and blue ellipses represent spin singlets .
the leftmost vertex @xmath156 prefers a down spin while the rightmost vertex @xmath157 prefers an up spin . ] because of the two different ways of pairing the spins , there are unpaired spins at vertices like @xmath156 and @xmath157 as shown in fig .
[ hchain ] .
note that because we have fixed a projection on the system , vertices @xmath156 and @xmath157 are different with different chiralities .
we can choose to have the spin to point up at vertex @xmath156 and point down at vertex @xmath157 .
in this way , it becomes natural that time reversal adds an extra @xmath152 at vertex @xmath156 and an extra @xmath153 at vertex @xmath157 .
in general , an unpaired spin always appears at vertices listed in eq .
[ t1 ] and [ t2 ] .
we choose to have the spin to point up at vertices in eq .
[ t1 ] and to point down at vertices in eq .
[ t2 ] . therefore ,
all these vertices transform under time reversal as @xmath72 .
there are other possible ways to ` attach haldane chains ' to realize the desired time reversal symmetry .
we summarize them in appendix [ fr_t ] , but the different choices do not affect our discussion in the following .
such a spin configuration not only fixes the time reversal symmetry action on the ground state , but also determines the transformation law of the deconfined anyonic excitations on the surface of the system .
for example , deconfined excitations of anyon type @xmath78 on the surface are created by adding open strings of type @xmath78 to the surface . at the point of excitation
, the wave function contains vertices like @xmath158 .
because @xmath78 carries a haldane chain with it ( while @xmath58 does not ) , at the vertex where the @xmath78 string ends , there is an extra spin @xmath127 degree of freedom which transform projectively under time reversal .
therefore , the anyonic excitation of type @xmath78 on the surface carries a projective representation of time reversal .
if the symmetry if not broken , the excitation gives rise to at least a two - fold degeneracy locally . in this section
, we discuss the details about the exactly solvable model with t - pfaffian surface state using the walker - wang construction .
readers not interested in the exact form of the hamiltonian can skip this section .
our construction here follows closely the strategy outlined in and we refer the reader there for further details .
the t - pfaffian walker - wang model is defined on a 3d trivalent lattice .
a trivalent 3d lattice can be obtained by splitting the vertices in a cubic lattice as shown in fig.[trivalent_lattice ] .
we have taken a particular projection of the 3d lattice onto the 2d plane .
each link in the lattice carries a @xmath159 dimensional spin degree of freedom , corresponding to the @xmath159 anyon types in t - pfaffian .
the hamiltonian contains vertex terms @xmath160 involving all three links ending at a particular vertex and plaquette terms @xmath161 involving links around a particular plaquette h=_v a_v + _ p b_p the vertex term enforces the fusion rules at each vertex by giving a higher energy to all disallowed vertices where the three strings do not fuse to the vacuum . the plaquette term is a sum over terms labeled by anyon type @xmath162 , @xmath163 , weighted by the quantum dimension of @xmath162 .
each @xmath164 can change the labels of links in a plaquette ( @xmath165 ) , and can also depend on the labels of adjoining links ( @xmath166 ) ( but can not change these ) . explicitly , the matrix element between a state with plaquette links ( @xmath165 ) and ( @xmath167 ) is l b^s , a, ...
,w_p , a, ... ,w=r^qb_q(r^cr_c)^*(r^qb_q)^*r^cr_c + _ ap_pq_qb_bc
+ _ cr_ru_ud_dv + _ vw_wa the intuition behind this complicated looking term is that it fuses in the loop @xmath162 to the skeleton of the plaquette using multiple @xmath29 moves , but in the process of doing so must use @xmath28 moves to temporarily displace certain links ( @xmath168 and @xmath169 in fig .
[ plaquette_term ] ) .
it is possible to check that all of these terms commute , and the the resulting ground state is a superposition of string - nets as given in eq .
[ ground_state ] . following the proof in
, we can see that the model has the bulk and surface deconfined excitations as desired . in order to realize time reversal symmetry on this model ,
we need to add two spin @xmath127 degrees of freedom to each link and put them into haldane chains or trivial chains along different strings . at each vertex
listed in eq .
[ t1 ] and [ t2 ] , there is an unpaired spin @xmath127 .
we add an up pointing magnetic field to this spin if it is at a vertex in eq .
[ t1 ] and a down pointing magnetic field if it is at one in eq .
the plaquette term should then be modified correspondingly to act only between the low energy configurations .
it can then be checked that the hamiltonian is indeed time reversal invariant . in order to add charge to this model
, we can use similar tricks and attach fractional charges to the end of each link .
when the links connect according to the fusion rules of the theory , the charges cancel at the vertex .
when excitations are created by end of strings , extra fractional charges are present at the position of the excitation .
there is one subtle point about this construction .
the walker - wang model as defined above is a spin model instead of a fermion model as one would expect for a 3d topological insulator .
the fermions in the bulk appear as gauge fermions coupled to a @xmath11 gauge field .
therefore , the model is topologically ordered not only on the surface but also in the bulk .
apart from the deconfined fermionic point excitations , there are also @xmath11 flux loop excitations in the bulk around which the gauge fermions pick up a @xmath46 phase factor .
a simple way to remove the topological order in the bulk and obtain a real fermionic model is to couple the above construction with a trivial fermionic insulator and condense the fermion pairs .
because the gauge fermion transform as @xmath72 under time reversal , it comes in two species @xmath170 and @xmath171 .
it is coupled to a fermionic insulator made also of @xmath72 fermions @xmath172 , @xmath173 .
one can then turn on the tunneling @xmath174 and condense the particle - hole pair .
the @xmath11 flux loops are then confined because they have nontrivial statistics with the condensate .
therefore , the bulk topological order is removed and one obtains a short range entangled fermionic model as desired .
the process of removing topological order by breaking charge conservation but not time reversal symmetry is realized in a simpler way in the pfaffian antisemion model @xcite . in this section
we show this model can be realized as the surface state of a @xmath0 and u(1 ) conserving bulk phase and discuss its properties compared to the free fermion ti surface .
the pfaffian - antisemion state is a product of the anti - semion state and the pfaffian state .
we label the anyons as @xmath175 .
the topological spins for the anyons are i & 1 & & i & & 1 & & i & + is & -i & & 1 & & -i & & 1 & + & & e^i/4 & & -e^i/4 & & -e^i/4 & & e^i/4 + s & & e^-i/4 & & -e^-i/4 & & -e^-i/4 & & e^-i/4 + & -1 & & -i & & -1 & & -i & + s & i & & -1 & & i & & -1 & + [ ts_spin ] the ising and anti - semion parts are both neutral while the @xmath1 part has charge @xmath176 .
time reversal symmetry exchanges the following pairs of anyons @xmath177 , @xmath178 , @xmath179 , @xmath180 , @xmath181 .
moreover , using the rules in section [ local_t ] , we can determine the local time reversal symmetry action on anyons which do not change type under time reversal : @xmath182 , @xmath78 , @xmath183 , @xmath79 , @xmath184 , @xmath59 , @xmath185 . because @xmath67 and @xmath68 map into each other under time reversal and braid with a @xmath46 , @xmath72 for their fusion product @xmath59 .
therefore , the pfaffian - antisemion state is the surface state of a @xmath72 topological insulator .
similarly , @xmath60 and @xmath186 are time reversal partners . in fusion channel @xmath182 , they have trivial mutual statistics , therefore , @xmath71 for @xmath182 ( also for @xmath183 ) . in fusion channel @xmath184
, they have mutual semionic statistics , therefore , @xmath72 for @xmath184 ( also for @xmath185 ) .
moreover , @xmath187 and @xmath188 are time reversal partners and have mutual semionic statistics .
therefore , @xmath72 for their fusion product @xmath79 .
finally , from the fusion of @xmath79 and @xmath59 into @xmath78 we find that @xmath71 for @xmath78 .
a walker - wang model can be written down explicitly for the pfaffian - antisemion state as well .
the model is defined on a 3d trivalent lattice similar to the t - pfaffian case while the dimension of the hilbert space on each link is doubled .
basis states on each link correspond to the 24 quasiparticles in the theory and the vertex term and plaquette term enforces the fusion and braiding rules . on the surface of the system , there are deconfined excitations corresponding to all 24 quasiparticles in the theory while in the bulk only the @xmath59 fermion is deconfined .
time reversal symmetry acts as complex conjugation , permutation of link basis states , and phase factors at each vertex .
the phase factors involved are such that vertices like @xmath189 transform as @xmath72 .
such vertex transformation laws can be understood as coming from the kramer degeneracy of @xmath79 , @xmath59 , @xmath184 , and @xmath185 .
a property of this model , which is different from t - pfaffian , is that if we condense @xmath182 , the topological order can be removed without breaking time reversal symmetry .
one may also observe that the fundamental @xmath190 vortices carry majorana zero modes . to see this first notice that naively , a charge @xmath36 condensate would entail vorticity in multiples of @xmath191 .
a vortex of strength @xmath190 would induce a berry phase of @xmath192 on the condensed particles .
however , this can be rectified by attaching a @xmath101 ( @xmath193 ) particle to the vortex ( to the anti vortex ) , which has the opposite braiding statistics that is , we can obtain the superconducting surface state of the free fermion ti starting from the pfaffian - antisemion state .
now we can take two copies of this state and see if it can be made trivial .
we label the anyons in the two copies as @xmath175 and @xmath194 .
first we set aside the anti - semion parts of the theory @xmath195 and @xmath196 and analyze the two pfaffian parts .
we can again condense the following set of composite bosonic particles without breaking time reversal or charge conservation : i_2_6,_2_6,i_6_2,_6_2,i_4_4,_0_0,_4_4 the particles that remain in the two pfaffian states include ( up to the condensed particles ) _ 1_3 , _ 1_7 ,
i_4,_0,_4 this is very similar to the t - pfaffian case except that the @xmath197 particle has topological spin @xmath198 and the @xmath109 particle has topological spin @xmath80 . in the resulting theory , these quantum dimension two particles split into two abelian particles @xmath199 , @xmath200 ( semions ) , which fuse to @xmath79 .
s_1 s_2 = _ 0 combined with the anti - semion part of the theory , the total theory after condensation is a product of two semion theories , two anti - semion theories and a trivial electron .
\{i , s_1 } \{i , s_2 } \{i , s } \{i , } \{i,_4 } the semion and anti - semions parts are all neutral .
only the electron is charged .
we can further condense @xmath201 and its time reversal partner @xmath202 without breaking any symmetry , which confines everything except the electron @xmath59 .
therefore , the pfaffian - antisemion state likely corresponds to a @xmath11 topological insulator . in order to check what happens between time reversal symmetry breaking domains
, we can couple the the pfaffian - antisemion surface state to 2d realizations of the pfaffian - antisemion state , which must break time reversal symmetry .
following similar analysis as above , we find that if we couple a 2d pfaffian - antisemion state on the left half of the plane , we can remove all topological order .
similarly , if we couple a time reversed 2d pfaffian - antisemion state on the right half of the plane , with time reversed coupling , we can also removed all topological order . between these two domains , a chiral edge state with @xmath203 , @xmath117 is left behind , which is expected for the free fermion topological insulator surface state .
we can directly show that one of the two t - pfaffians can be connected to the pfaffian - antisemion state , i.e .
is surface equivalent to it .
again , the procedure is not constructive - so we do nt know which of the two it is .
we do not make any assumption about the classification of interacting topological insulators ( ref . ) but utilize a related construction .
consider combining the t - pfaffian state with pfaffian - antisemion state and attempting to remove all the topological order while preserving both @xmath134 and charge @xmath1 symmetry . to help this process
we introduce a z@xmath204 gauge theory topological order of cooper pairs , where the gauge charges @xmath205 @xmath206 carry charges @xmath207 , and the gauge fluxes @xmath208 , where @xmath209 are exchanged under time reversal symmetry @xmath210 under @xmath134 .
@xmath211 maps to itself under time reversal and we choose it to transform as a kramer doublet .
consider condensing @xmath212 and its time reversed conjugate @xmath213 .
note , here the first particle is in the pfaffian - antisemion theory while the second is in the t - pfaffian ( with tilde ) .
these condensates are self and mutual bosons with charge 0 and preserve time reversal symmetry .
the last is particularly crucial since in the absence of the @xmath214 factors , the square of these condensates @xmath215 would break @xmath134 symmetry since one of these bosons is a kramers doublet .
however here it appears as @xmath216 which is a total time reversal singlet and can be condensed . as @xmath212 is a strength-@xmath73 flux of the whole theory , all the surviving excitations after this condensation have integer electric charge in units of @xmath17 .
we can separate all the quasiparticles after this condensate into two sets , a charge neutral set and a charge @xmath17 set which are related to each other through combination with an electron .
the neutral set form a closed modular topological theory @xmath217 which is @xmath134 symmetric ( note , @xmath134 symmetry does not interchange particles differing by an electron since they carry different charges)@xcite and the whole theory can be written as @xmath218 , where the electric charge is only carried by the electron @xmath219 . therefore
, the t - pfaffian state and the pfaffian - antisemion state is surface equivalent up to either the surface of a 3d bosonic spt of neutral boson protected by @xmath220 or a 2d @xmath134 symmetric topological order .
again , one can exclude the 3-fermion state since the combined theory has no nontrivial edge states even when realized in 2d .
so the question is whether they differ by a bosonic spt which is etmt ( both e , m transforming as kramers doublets ) .
now , since the two t - pfaffians differ by precisely this theory , one of them is equivalent to the pfaffian - antisemion state .
we have discussed the possibility of a 3d topological insulator with a symmetric gapped surface state and non - abelian surface topological order .
we constructed a model for a 3d topological insulator , with magnetoelectric response @xmath3 , with a surface state given by t - pfaffian topological order .
we find that the symmetry transformation on the t - pfaffian state can take two different forms .
one of them is consistent with being the surface state of the free fermion ti while the other differ from the free fermion ti by a neutral bosonic topological superconductor .
one remaining question is which specific t - pfaffian state is connected to the free fermion surface state .
we can not yet answer this question due to the lack of a simple way to smoothly connect the t - pfaffian state to the superconducting surface state of the free fermion ti , and is left for future work .
we also constructed an exactly soluble 3d lattice model for a somewhat more complicated topological order , the pfaffian - antisemion theory@xcite with twice as many particles , which can be smoothly connected to the superconducting surface state .
these finds are consistent with the classification result obtained in .
xc is supported by the miller institute for basic research in science at uc berkeley .
xc thanks discussions with p. bonderson , x.l .
qi , m. fisher , t. senthil , d. potter , c. wang , f. burnell , z.h .
av thanks ehud altman , parsa bonderson and particularly t. senthil for several insightful discussions , and p. bonderson , max metlitski and t. senthil for discussing their results prior to publication .
av is supported by aro muri grant w911-nf-12 - 0461 .
in order to see , with broken @xmath1 symmetry whether the t - pfaffian theory can be realized in two dimension with time reversal symmetry , we try to gauge the @xmath11 fermion parity symmetry in the theory and examine whether the resulting theory can be time reversal invariant or not .
there are different ways ( at least 16 ) to consistently gauge the @xmath11 fermion parity symmetry and we need to consider them all .
one obvious way to gauge the @xmath11 fermion parity in the t - pfaffian theory is into the full @xmath221 theory .
the topological spins of all @xmath222 particles are listed below i & 1 & e^i & i & -e^i & 1 & -e^i & i & e^i + & e^-i & 1 & e^i & -1 & e^i & -1 & e^i & 1 + & -1 & -e^i & -i & e^i & -1 & e^i & -i & -e^i + the braiding of the @xmath223 particle correctly measures the ( fractional ) @xmath11 quantum number associated with the original anyons .
therefore , @xmath223 is the @xmath11 flux and the full table is obtained by combining @xmath223 with all the original anyons .
this theory is obviously not time reversal invariant
. there are ( at least ) 15 other different ways to gauge the @xmath11 fermion parity symmetry in this theory .
they can be obtained by combining one of kitaev s 16 fold way to gauge free fermion theory together with the full @xmath221 theory and condense the fermion - fermion pair .
for example , if we consider the @xmath110th one in kitaev s 16 fold way , the toric code model with @xmath40 , @xmath224 , @xmath225 , @xmath226 , then the anyon types in the resulting theory comes in two sets : the set of anyons in the t - pfaffian theory and the set of anyons in @xmath227 . and topological spins in the resulting theory are ( here we are still denoting the anyons with the label in @xmath221 but half of them are actually combined with @xmath225 in the toric code theory . )
i & 1 & e^i & i & -e^i & 1 & -e^i & i & e^i + & e^-i & 1 & e^i & -1 & e^i & -1 & e^i & 1 + & -1 & -e^i & -i & e^i & -1 & e^i & -i & -e^i + which is the same as the @xmath221 theory , as we expected .
if we consider an even number element ( @xmath228-th ) in the 16 fold way , with anyons @xmath40 , @xmath224 , @xmath225 , @xmath226 and topological spins @xmath45 , @xmath229 , @xmath229 , @xmath46 , then the anyon content in the resulting theory is similar to the previous case and the topological spins are ( with slight abuse of notation for anyon label ) i & 1 & e^i(+1 ) & i & -e^i(+1 ) + & e^i(-1 ) & 1 & e^i(+3 ) & -1 + & -1 & -e^i(+1 ) & -i & e^i(+1 ) + & 4 & 5 & 6 & 7 + i & 1 & -e^i(+1 ) & i & e^i(+1 ) + & e^i(+7 ) & -1 & e^i(+3 ) & 1 + & -1 & e^i(+1 ) & -i & -e^i(+1 ) + the theory can not be time reversal invariant for any even @xmath228 .
now let s check the case for odd @xmath228 . with odd @xmath228
, the gauged free fermion theory has a nonabelian @xmath11 flux @xmath230 with topological spin @xmath229 .
the total theory is obtained by combining the ising@xmath231 theory , the @xmath38 theory and the @xmath228-th gauged free fermion theory and condensing @xmath232 , where the subscript @xmath233 denotes the @xmath1 charge .
the resulting theory contains the following anyons : llll ii_0=_4 & ii_2=_6 & ii_4=_0 & ii_6=_2 + i_0=i_4 & i_2=i_6 & i_4=i_0 & i_6=i_2 + i_1=_5 & i_3=_7 & i_5=_1 & i_7=_3 + i_1=_5 & i_3=_7 & i_5=_1 & i_7=_3 + _ 0=_4 & _ 2=_6 let us use the shorthand notation llll i_0 & i_2 & i_4 & i_6 + _ 0 & _ 2 & _ 4 & _ 6 + ^a_1 & ^a_3 & ^a_5 & ^a_7 + ^b_1 & ^b_3 & ^b_5 & ^b_7 + _ 0 & _ 2 the @xmath55 and @xmath56 particles have quantum dimension @xmath45 , the @xmath57 particles have quantum dimension @xmath44 and the @xmath234 particles have quantum dimension @xmath73 .
the @xmath235 particles could split into two quantum dimension @xmath45 particles . here _ 0_0
= i_0 + i_4 + _ 0 + _ 4 @xmath236 fuses with itself into four different particles , therefore @xmath236 does not split .
neither does @xmath237 .
the topological spins are llll 1 & i & 1 & i + -1 & -i & -1 & -i + 1 & -1 & -1 & 1 + e^i & -e^i & -e^i & e^i + e^i & e^i when @xmath238 , the anyon theory could potentially be time reversal invariant with topological spins llll 1 & i & 1 & i + -1 & -i & -1 & -i + 1 & -1 & -1 & 1 + 1 & -1 & -1 & 1 + e^-i & e^i the fusion and braiding all follow from the parent @xmath239 theory . under time reversal @xmath236 would map into @xmath237 and vice verse .
moreover , time reversal maps between the following pairs of particles i_2 _ 2 , i_6 _ 6 , ^b_1 ^b_7 , ^b_3 ^b_5 , _ 0 _ 2 we can check that this time reversal symmetry action is consistent with the fusion rule of the theory .
this is in contrast to the semion - fermion theory where one of the gauged theories does have time reversal pairs of topological spins but time reversal symmetry action is not consistent with the fusion rule . in that case
, the gauged theory is obtained from @xmath240 by condensing the @xmath241 pair .
the two dimensional particle @xmath242 needs to split into two one dimensional particles in order to make the theory unitary which causes the inconsistency between time reversal symmetry action and the fusion rule .
however , in the gauged t - pfaffian case , the @xmath243 particle does not split and time reversal can be consistent with fusion rule .
so now we need more powerful tools to determine whether t - pfaffian or its gauged version can be realized in 2d with time reversal .
the gauged t - pfaffian theory is a bosonic nonchiral theory . assuming it has time reversal symmetry , so it should either be ( 1 ) realizable in 2d ( 2 ) realizable on the surface of a 3d bosonic topological superconductor ( the @xmath11 class within group cohomology ) .
note , @xmath59 is a time reversal doublet since it is composed by fusing a pair of particles @xmath68 , @xmath67 which are mutual semions and go into each other under @xmath0 symmetry .
this implies @xmath78 is a kramers doublet too , and in that case we can not condense @xmath78 .
however , we can combine this theory with the changeless t - pfaffian ( i.e. having broken u(1 ) symmetry ) , and condense the product of the @xmath59 particles ( @xmath244 ) in the two theories .
this boson can be condensed and confines the flux excitations in the gauged theory resulting in two copies of t - pfaffian , which can be confined .
thus , the changeless t - pfaffian can be converted into a confined sc without topological order , by combining it with this state .
hence , assuming t symmetry of the gauged theory , the changeless t - pfaffian is either trivial ( realizable in 2d with t symmetry ) or equivalent to the surface of a 3d @xmath0 symmetric bosonic spt phase .
in this section , we show how to obtain the rules for determining the local action of time reversal symmetry given in section [ local_t ] from the walker - wang construction .
we apply the procedure to the case of t - pfaffian and find which anyons transform as @xmath72 . we discuss how various gauge choices in the problem can affect the solution .
@xmath74 at each vertex is then obtained by ( t^2)^ij_k=(^ij_k)^*^|i|j_|k which is equal to the combination of @xmath74 for @xmath80 , @xmath81 and @xmath52 .
( t^2)^ij_k = t^2_it^2_jt^2_k of course , for anyon types @xmath80 which change under time reversal , @xmath74 is not well defined .
however , we have the constraint that ( t^2_i)^*=t^2_|i in particular , if @xmath246 , @xmath247 . from these equations , we can derive the rules given in section [ local_t ] .
moreover , we shall see that the arbitrariness of @xmath74 for @xmath80 which is not equal to @xmath83 is taken into account naturally in these equations . when @xmath80 , @xmath81 and @xmath52 are all invariant under time reversal and @xmath80 , @xmath81 fuse into @xmath52 , ( t^2)^ij_k=(^ij_k)^*^ij_k=1 therefore , t^2_k = t^2_it^2_j as given in the first rule in section [ local_t ] .
next , when @xmath248 and @xmath80 and @xmath83 fuse into @xmath52 ( @xmath249 ) ( r^i|i_k)^*=r^|ii_k therefore , ( t^2)^i|i_k=(^i|i_k)^*^|ii_k = r^|ii_kr^i|i_k = s^i|i_k where @xmath250 is the phase factor coming from a full braid of @xmath80 around @xmath83 in fusion channel @xmath52 .
moreover , ( t^2)^i|i_k = t^2_it^2_|it^2_k = t^2_k therefore , t^2_k = s^i|i_k as given in the second rule in section [ local_t ] .
in particular in the case of t - pfaffian , solving the equations in eq . [ fr_t_eq ] with the @xmath29 and @xmath28 matrices given in section [ wwa ] , we obtain a set of solutions for @xmath150 .
pick one possible solution and we can determine the @xmath74 transformation law for each vertex @xmath251 from ( t^2)^ij_k=(^ij_k)^*^|i|j_|k we find that the following vertices have @xmath72 the @xmath74 of each vertex is determined by the @xmath74 of all three anyons involved .
( t^2)^ij_k = t^2_it^2_jt^2_k from this , we find that the possible set of kramer doublet anyons has two possibilities under this change , the @xmath151 s change by ^ij_k ^ij^ij_k^|i|j_|k this does not affect the value of @xmath254 for each vertex but can change the value of @xmath150 to be anything consistent with the @xmath254 value . using this degree of freedom , we can make @xmath151 s to be @xmath80 for vertices in eq .
[ t1 ] , @xmath198 for vertices in eq .
[ t2 ] and @xmath45 for all other vertices .
the @xmath151 s satisfying eq . [ fr_t_eq ] are not unique .
in particular , if they change by ^ij_k=^ij_k then obviously , the equations for @xmath29 and @xmath28 are still satisfied
. such a change would lead to a change in @xmath255 .
if @xmath150 changes by @xmath256 , then @xmath254 changes by @xmath257 .
correspondingly such a change lead to the change in @xmath258 as t^2_i t^2_i if @xmath246 , then @xmath258 does not change .
if @xmath248 , @xmath258 does change .
however , @xmath258 is not well defined for @xmath248 and the degree of freedom given by @xmath259 reflects exactly this arbitrariness .
note that with arbitrary @xmath260 , we have ( t^2_i)^*=t^2_|i | the surfaces of three dimensional topological insulators ( 3d tis ) are generally described as dirac metals , with a single dirac cone .
it was previously believed that a gapped surface implied breaking of either time reversal @xmath0 or @xmath1 charge conservation symmetry .
here we discuss a novel possibility in the presence of interactions , a surface phase that preserves all symmetries but is nevertheless gapped and insulating .
then the surface must develop topological order of a kind that can not be realized in a 2d system with the same symmetries .
we discuss candidate surface states - non - abelian quantum hall states which , when realized in 2d , have @xmath2 and hence break @xmath0 symmetry .
however , by constructing an exactly soluble 3d lattice model , we show they can be realized as @xmath0 symmetric surface states .
the corresponding 3d phases are confined , and have @xmath3 magnetoelectric response .
two candidate states have the same 12 particle topological order , the ( read - moore ) pfaffian state with the neutral sector reversed , which we term t - pfaffian topological order , but differ in their @xmath0 transformation .
although we are unable to connect either of these states directly to the superconducting ti surface , we argue that one of them describes the 3d ti surface , while the other differs from it by a bosonic topological phase .
we also discuss the 24 particle pfaffian - antisemion topological order ( which can be connected to the superconducting ti surface ) and demonstrate that it can be realized as a @xmath0 symmetric surface state . |
the casimir effect has in recent years attracted considerable interest . for reviews on this topic see , for example , plunien _ _ et al.__@xcite , mostepanenko and trunov@xcite , and milton @xcite .
the problem has been rather difficult to solve , once curved boundaries are present , because of divergences in the formalism and insufficient contact with experiments . in this paper
we will consider a compact ball of radius a , permittivity @xmath4 and permeability @xmath5 , situated in an infinite medium whose corresponding material constants are @xmath6 and @xmath7
. we shall take @xmath8 [ 1 ] so that the photon velocity is everywhere equal to @xmath9 .
the relativistic condition ( 1 ) was , in the context of the casimir effect , to our knowledge first introduced by brevik and kolbenstvedt @xcite .
the reason why this particular condition was imposed was that when calculating the casimir surface force density on the sphere , in this paper called @xmath10 , one avoids thereby the subtraction of a contact term : the contact term turns out to be zero just as in the case of a singular spherical conducting sphere in a vacuum @xcite .
later , this kind of medium has been considered from various viewpoints [ 7 - 19 ] .
a common feature of most of these papers is the use of macroscopic electrodynamics of continuous media .
the main purpose of the present paper is to consider the casimir effect for a compact ball satisfying the condition ( 1 ) , at finite temperatures .
finite temperature casimir theory has been considered previously by several authors , e.g. , lifshitz @xcite , fierz @xcite , sauer @xcite , mehra @xcite , brown and maclay @xcite , milton _ et al . _
@xcite , revzen _ et al . _
@xcite , and klich @xcite .
also the very recent extensive treatise of feinberg _ et al .
_ @xcite ought to be mentioned , since it describes in detail the characteristic properties of the high temperature limit , and contains also an extensive reference list .
we start in the next section by considering , as a first step , the @xmath11 theory , and demonstrate the equivalence between the green function method and the mode summation method .
the equivalence has of course to hold if the theory is to make sense physically , but the equivalence is not so easy to see by a mere inspection of the mathematics .
a detailed calculation is desirable . in section 3
we generalize the casimir energy expression obtained in @xcite to the fourth order in the diluteness parameter @xmath12 ; cf .
( 15 ) and the definition in eq .
( 7 ) . in section 4
we consider the casimir free energy @xmath13 at finite @xmath14 , emphasizing the difference between @xmath13 and the casimir energy @xmath15 , and give an order of magnitude estimate of @xmath13 in the case of a dilute medium endowed with a `` square '' dispersion relation implying a sharp high frequence cutoff at imaginary frequency @xmath16 .
the presence of a dispersion - induced strong , negative , cutoff - dependent contribution to the casimir energy ( or free energy , for finite @xmath14 ) , seems first to have been pointed out by candelas @xcite , for the case of a perfectly conducting shell . at high temperatures we find @xmath13 to be negative , decreasing linearly with @xmath14
this agrees with results found previously in @xcite and @xcite .
the casimir energy @xmath15 itself tends to zero at high temperatures .
in this section we consider first the casimir surface force density @xmath10 at zero temperature , for a nondispersive medium .
there are in principle two different methods for calculating @xmath10 ( more precisely , the part of @xmath10 that is independent of the high frequency cutoff ) : the first method is to make use of the green functions .
there are two scalar green functions , corresponding to two different electromagnetic modes .
again , this method can be carried out in one of two different ways : 1 .
first , one can evaluate @xmath10 as the difference between the radial maxwell stress tensor components on the outside and inside of the surface @xmath17 .
we have to subtract off the volume terms in the scalar green functions , so that only the boundary dependent terms are left .
this regularization ensures that the surface force goes to zero when @xmath18 , which is as we expect on physical grounds .
also , this regularization permits us to calculate the casimir energy @xmath19 once @xmath10 is known , by the formula @xmath20 .
( at @xmath11 , @xmath19 is the same as the free energy @xmath21 . ) 2 .
alternatively , one can calculate the energy @xmath19 by integrating the maxwell energy density over the volume , retaining , as above , only the boundary dependent terms in the green functions .
the outcome of these two approaches are of course in agreement .
the green function method was first employed by milton , for the case of an ordinary dielectric ball @xcite , and was later applied to the case of an @xmath1 ball in ref .
@xcite .
the result of the calculation can be given as follows .
after a complex frequency rotation @xmath22 we can express all physical quantities as integrals over frequencies @xmath23 along the imaginary frequency axis .
the cutoff dependence is omitted ( it disappears in the regularization procedure ) .
we define riccati - bessel functions @xmath24 and @xmath25 by ( @xmath26 @xmath27 [ 2 ] corresponding to the wronskian @xmath28 . here
@xmath29 ; @xmath30 and @xmath31 are modified bessel functions .
we introduce two dispersion functions @xmath32and @xmath33corresponding to the te and tm modes : @xmath34 [ 4 ] the casimir energy as derived in @xcite on the basis of the green function can be written as @xmath35'.\ ] ] [ 5 ] here @xmath36 , @xmath37 ; the time difference @xmath38 has been euclidean rotated , @xmath39 , and @xmath40 is the time - splitting parameter .
moreover , we have defined the function @xmath41 as @xmath42'$ ] [ in ref @xcite we used the symbols @xmath43 and @xmath44 ; the relations between these quantities and those defined by ( 3 ) and ( 4 ) are @xmath45 and @xmath46 . ]
consider next the second method , which consists in summing over the modes by means of contour integration .
this method was recently used in ref .
@xcite , and is in turn based on the technique developed by lambiase and nesterenko @xcite and nesterenko and pirozhenko @xcite .
one gets ( when we supply the same parameter @xmath47 as above ) @xmath48.\ ] ] [ 6 ] introducing the parameter @xmath12 by @xmath49 [ 7 ] the expression ( 6 ) can alternatively be written as @xmath50.\ ] ] [ 8 ] the two methods thus lead to two expressions for the casimir energy , given by eqs .
( [ eq:5 ] ) and ( [ eq:6 ] ) .
the expressions , of course , have to be equal , but to see the equality is not quite trivial by mere inspection .
the equality is tantamount to the equality @xmath51'=
-\frac{(\mu_1-\mu_2)^2}{4\mu_1\mu_2 } \left[\lambda_l^2(x)\right]'.\ ] ] [ 9 ] now it turns out that this equality is not so difficult to prove after all if we insert the expressions ( 3 ) and ( 4 ) on the left hand side and take into account the above mentioned wronskian between @xmath24 and @xmath25 .
actually , the expression ( 6 ) was also made use of in the green - function calculation of milton and ng @xcite .
before leaving the zero temperature theory let us make a brief remark on the nondispersive dilute ball , meaning that @xmath52 [ 10 ] the most practical expression to make use of for the casimir energy , is that of eq.([eq:8 ] ) .
we expand the logarithm to fourth order in @xmath12 : @xmath53 [ 11 ] and make the following steps : 1 .
put @xmath54 ; 2 . perform a partial integration in eq.(8 ) ; 3 .
interchange summation and integration signs .
then , omitting the @xmath55 terms in ( [ eq:11 ] ) , @xmath56 [ 12 ] which can be processed further using the uniform asymptotic expansion for @xmath24 and @xmath25 up to @xmath57 , in the same way as in refs .
there is one single divergent term in eq .
( 12 ) , which can be regularized by means of the riemann zeta function , @xmath58 .
the only formula needed in practice is @xmath59 [ 13 ] moreover , we use the summations @xmath60 , @xmath61 , as well as the integral @xmath62 [ 14 ] ( @xmath63 is the beta function ) , to get @xmath64\\ & -\frac{63 \xi^4}{16384 a } \left[\left(\half\pi^2 - 4\right)-\frac{351}{1792}\left(\frac { 1 } { 6}\pi^4 -16 \right)\right ] .
\end{split}\ ] ] [ 15 ] this expression generalizes that of equation ( 3.10 ) in @xcite up to the fourth order in @xmath12 .
numerically , the expressions between square parentheses in ( 15 ) are respectively 1.05966 and 0.88880 . the fourth order contribution to the casimir energy
is thus weak and negative , corresponding to an attractive force component . in connection with this calculation
, we wish to emphasise two points .
first , we may avoid the time - splitting parameter @xmath47 from the beginning , using zeta function regularization ( here the riemann function ) instead .
secondly , the convergence parameter @xmath65 that was for mathematical reasons introduced in @xcite , can simply be omitted . at least from a pragmatic point of view
, the nondispersive medium becomes quite simple .
we now take into account dispersive properties of the two media , meaning that @xmath66 , @xmath67 .
we shall still assume that @xmath68 , so that the velocity of photons in either medium is equal to the velocity of light .
after frequency rotation , along the imaginary frequency axis , @xmath69 , @xmath70 .
some care ought to be taken when considering the dispersive formalism , since frequency derivatives of the material constants are no longer ignorable straightaway .
let us first consider the @xmath11 total zero - point energy , which we shall call @xmath71 , of the inner ( i ) and the outer ( ii ) regions .
this energy is simply given by eq .
( 6 ) , which we repeat here with a slightly generalized notation : @xmath72.\ ] ] [ 16 ] the reason for this equality is that the two dispersion functions @xmath73 and @xmath74 , as given by the dispersive generalization of eqs .
( 3 ) and ( 4 ) , retain their physical meaning also in the presence of dispersion .
thus the te and tm modes are still determined by @xmath75 and @xmath76 respectively . the expression ( 16 )
needs to be regularized : we subtract off the zero - point energy @xmath77 corresponding to the limit @xmath78 ; cf . @xcite . as @xmath79 and @xmath80 for large @xmath81
, we get from ( 3 ) and ( 4 ) @xmath82 [ 17 ] the casimir energy @xmath83 becomes accordingly , at @xmath11 , @xmath84.\ ] ] [ 18 ] observing that @xmath85 , @xmath86 we find ( when reverting to the notation @xmath87 instead of @xmath88 ) that @xmath89,\ ] ] [ 19 ] which finally means that , at @xmath11 , @xmath90.\ ] ] [ 20 ] it is rather remarkable that this simple formula continues to hold even in the presence of arbitrary frequency dispersion . since dispersion implies that we introduce a physically based high frequency cutoff , we expect that there is no longer any need for a time splitting parameter .
we have therefore put @xmath54 in the expression ( 20 ) .
we will henceforth restrict ourselves to the first ( i. e. , the second order ) term in the expansion ( 11 ) . in this section
we adopt the simplest imaginable dispersion relation ( a `` square '' form ) : we let the inner medium ( a ball ) correspond to @xmath91 [ 21 ] whereas the outer medium is a vacuum .
it is to be noted here that the general thermodynamic law saying that the susceptibility has to decrease monotonically along the positive imaginary frequency axis @xcite , is satisfied by this curve for @xmath92 if it is given a small negative slope for @xmath93 .
we still assume dilute media , @xmath94 , as above .
consider now the case of finite temperatures .
since the thermal green function is periodic in imaginary time , we can replace the fourier transform in the @xmath11 theory with fourier series in imaginary time .
the transition to finite temperature theory is accomplished by means of a discretization of the frequencies , @xmath95 [ 22 ] where @xmath96 is an integer and @xmath97 ( we put @xmath98 ) .
the rule for going from frequency integral to sum over matsubara frequencies is @xmath99 [ 23 ] here @xmath100 is a nondimensional temperature , and the prime on the summation sign in ( 23 ) means that the @xmath101 term is counted with half weight .
since @xmath102 can be written as @xmath103 , we can express the casimir free energy at finite temperatures as @xmath104 [ 24 ] to @xmath105 .
the corresponding zero temperature expression , still to be called @xmath19 , is given by the first term to the right in eq .
( 12 ) .
the following delicate point ought to be noted . we know that the mean energy per mode of the electromagnetic field is @xmath106 .
the function @xmath107 is to replace @xmath108 , when we construct the general contour integral for the casimir energy @xmath15 at finite temperatures .
it is the function @xmath109 which is responsible for the occurrence of the matsubara frequencies along the imaginary frequency axis .
now , our formalism implies morover a transition from the energy @xmath15 to the free energy @xmath13 .
this is accomplished by means of a partial integration , making use of the relations @xmath110 for this reason we have changed the energy symbol from e to @xmath21 in eq .
( 24 ) . at @xmath11 , obviously @xmath111 .
we thank i. klich ( personal communication ) for helpful comments regarding this point .
let us in this context also recall how the surface force density @xmath10 is calculated . generally ,
as mentioned in section 2 , @xmath10 can be found as the difference between the radial maxwell stress tensor components on the outside and inside of the surface @xmath17 .
the stress components are in turn constructed from the two - point functions like @xmath112 when @xmath113 and @xmath114 are close together , at a given temperature @xmath14 .
the appropriate energy function is found by integrating @xmath115 over @xmath0 , from initial position @xmath116 , to the final position where the radius of the sphere terminates at the fixed value @xmath0 .
this process is to take place at constant temperature .
the appropriate energy function is accordingly the _ free energy
_ @xmath13 .
thus we see that , whereas in force considerations at finite temperatures one is lead naturally to the free energy , while starting from energy considerations one has instead to make use of thermodynamical considerations to calculate @xmath13 .
mathematically , the case @xmath101 requires special attention . for this purpose
we observe the approximate expression for @xmath24 and @xmath25 at low @xmath81 : @xmath117 [ 25 ] from this it follows that @xmath118 , @xmath119 , implying that the contribution to ( 24 ) from @xmath101 is @xmath120 [ 26 ] we add the contribution from @xmath121 , restricting ourselves to the dominant term in the uniform asymptotic expansion of @xmath122 @xcite : @xmath123 .\ ] ] [ 27 ] thereby @xmath124.\ ] ] [ 28 ] we have here truncated the sum over @xmath125 at @xmath126 in order to avoid divergences .
it is easy to give an order - of - magnitude estimate of @xmath127 : our dispersion relation ( 21 ) implies that photons having frequencies @xmath128 higher than @xmath129 do not `` see '' the sphere at all .
a photon of limiting frequency @xmath129 just touching the surface of the sphere has an angular momentum equal to @xmath130 .
this type of argument has repeatedly been used in casimir calculations @xcite .
one might put @xmath131 where @xmath9 is a coefficient of order unity .
however , in view if simplicity , and since we are able to give only an estimate of the order of magnitude anyhow , we put henceforth @xmath132 .
as upper limit in the summation we thus take @xmath133 [ 29 ] now taking into account the exact summation formula derived in eq.(a3 ) in @xcite ( with @xmath134 an arbitrary constant ) : @xmath135 [ 30 ] we can write eq . ( 28 ) as @xmath136;\ ] ] [ 31 ] the first term on the right hand side of ( 30 ) drops out .
this expression can be calculated , with @xmath137 as an input parameter , to show how @xmath13 varies with @xmath100 .
it is of interest first to examine the case of low temperatures .
let us assume that @xmath138 [ 32 ] implying that @xmath139 .
then the first term in ( 31 ) dominates , and can be replaced by unity .
we obtain approximately @xmath140 [ 33 ] this negative expression establishes a constant low - temperature plateau for the casimir free energy .
it is of interest to compare this result with that obtained from the slightly more complicated though physically more realistic lorentz ( or sellmeir ) dispersion relation . the latter , when taken only along the imaginary frequency axis , can be written as @xmath141 [ 34 ] where @xmath142 is the zero frequency magnetic susceptibility .
( the damping term is omitted . ) here @xmath143 represents a `` soft '' cutoff .
this case was considered in @xcite , for the zero - temperature case . for a dilute medium , @xmath144 , the result of the calculation in @xcite
can be written @xmath145 [ 35 ] this formula assumes mathematically , strictly speaking , that @xmath146 in addition to @xmath147 .
the first condition is however not very stringent ; it turns out numerically that ( 35 ) holds to better than 1 percent accuracy even when @xmath137 is as low as about 4 .
we thus see that the dominant first term in eq .
( 35 ) agrees qualitatively well with eq .
( 33 ) , the latter obtained on basis of the `` square '' dispersion relation ( 21 ) .
the latter expression is @xmath148 of the lorentz - based expression @xmath149 .
this is a satisfactory result , in view of the semi - quantitative agreement between the meaning of the parameters @xmath137 occuring in ( 16 ) and in ( 34 ) .
the second term in eq . ( 35 ) , @xmath150 , is the characteristic positive casimir energy ( corresponding to a repulsive force component ) for a _ nondispersive _ dilute ball satisfying the condition @xmath151 , at zero temperature .
, for instance , @xcite .
the other limiting case of interest is that of high temperatures .
let us assume that @xmath152 [ 36 ] implying that @xmath153 for all values of @xmath125 in eq .
( 31 ) . making use of @xmath154 and @xmath155 for small @xmath156 , we obtain as leading term @xmath157 [ 37 ]
this expression is independent of @xmath0 .
the high - temperature casimir free energy , still negative , thus decreases proportionally with @xmath100 .
this is qualitatively in agreement with @xcite and @xcite .
figures 1 - 3 show how the nondimensional reduced casimir free energy @xmath158 , defined by @xmath159 [ 38 ] varies with @xmath100 , in the cases when @xmath160 .
the low - temperature plateaus evidently lie at @xmath161 .
these figures are given on a linear scale , thus exhibiting the linear behaviour predicted by eq .
( 37 ) at high temperatures .
it is of interest to check the accuracy of ( 37 ) in cases when the condition ( 36 ) is reasonably well , but not extremely well , satisfied .
let us choose @xmath162 as an example .
from the expression @xmath163 [ 39 ] we obtain , taking into account that @xmath164 , that @xmath165 the full series solution yields the number -20.047154 .
the approximate formula ( 39 ) is thus in this case extremely accurate .
choosing @xmath166 , we get from eq . ( 39 ) that @xmath167 , whereas the series ( 31 ) yields -2.091 . even in this case , where the condition ( 36 ) obviously is broken , the error in the formula ( 39 ) is thus only about 4 per cent . generally speaking , at high temperatures
when the diluteness of the medium becomes more pronounced , the casimir free energy becomes small , as we would expect . in connection with the temperature dependence of the free energy the following physical argument should be noticed . from the dispersion relations ( 21 ) or ( 34 )
we would expect that the influence from the dispersion effect becomes important when the temperature becomes so high that the most significant frequencies in the thermal radiation field are of the same order as @xmath129 . and as most significant thermal frequency , it is natural to take the value @xmath168 corresponding to the _ maximum _ of the blackbody distribution . from wien s dispacement law we know that @xmath169 .
thus , we expect that dispersion becomes significant when @xmath170 which , in view of the definition @xmath171 , means that @xmath172 .
now , in figs . 1 - 3
the linear scale chosen for the temperature implies that the intervals covered on the ordinate axes are small , and the mentioned effect becomes hidden ( the `` shoulder '' in each diagram appears to lie at a lower value of @xmath100 ) .
the effect becomes however clear if we replace the linear temperature scale by a logarithmic scale .
figure 4 shows , as an example , such a diagram for the case @xmath173 .
then the ordinate interval becomes considerably larger , and we see that the prediction @xmath174 for the position of the " shoulder turns out to be reasonable .
( in the diagram we have drawn the line @xmath175 to emphasize the transitional region . )
the same behaviour was previously found to hold in @xcite .
we note in passing that the high temperature entropy @xmath176 turns out to be constant : from eq .
( 37 ) we get @xmath177 [ 40 ]
instead of the formal divergence encountered in the previous section and the necessity to truncate the sum over @xmath125 in eq .
( 28 ) at an upper limit @xmath126 , it is of interest to consider a mathematically more involved but physically more correct dispersion relation , that leads to a finite result for the free energy even after summing over all @xmath125 up to infinity .
let us consider the following dispersion relation , being essentially of the lorentz form but augmented by a term in the denominator that describes spatial dispersion : @xmath178 [ 41 ] @xmath142 being a constant .
as in the previous section we assume that the ball of radius @xmath0 is dilute ( @xmath179 ) , and that the exterior region is a vacuum .
the spatial dispersion term @xmath180 ensures convergence when summing over all @xmath125 .
since @xmath181 according to ( [ eq:7 ] ) , we obtain to first order in @xmath142 , at finite temperatures , @xmath182 [ 42 ] instead of the expression ( 28 ) we now get @xmath183,\ ] ] [ 43 ] where we have introduced the symbol @xmath184 .
we resolve the last term into partial fractions , and introduce the symbol @xmath185 : @xmath186 [ 44 ] then , if @xmath134 and @xmath187 are arbitrary constants , we get @xmath188 [ 45 ] from appendix a in @xcite we write down the expressions for @xmath189 ; @xmath190 : @xmath191 [ 46 ] @xmath192 [ 47 ] ( @xmath193 was given in eq .
we define , in analogy with eq .
( 38 ) , the nondimensional reduced free energy to be @xmath194 [ 48 ] and obtain after some manipulations the following expression : @xmath195 \\ & - \frac{\nu^5}{t^4 ( x_0 ^ 2-l-\frac{1}{4})^3}\left [ s ( \frac{\sqrt{x_0 ^ 2+l^2}}{t};2 ) + 2s(\frac{\nu}{t};2 ) \right ] \\ & + \left .
\frac{\nu^5}{t^6 ( x_0 ^ 2-l-\frac{1}{4})^2}s ( \frac{\nu}{t};3 ) \right\}. \end{split}\ ] ] [ 49 ] this expression is calculated numerically for two of the cases above , viz . @xmath196 .
the results are shown in fig .
it is seen that the curves have roughly the same form as in figs . 1 - 2 .
there is a definite low - temperature plateau , and for high values of @xmath100 the curves tend towards a linear form . in any case , whatever we adopt a square dispersion relation or a spatial - dispersion relation , the sensitivity with respect to increasing values of @xmath100 is most pronounced when @xmath137 is smallest .
the limiting values for @xmath197 in fig .
5 are in agreement with those of figs . 1 and 2 to within about 70 per cent .
this is the order of magnitude - agreement that we might expect , since the parameter @xmath198 made use of in eq .
( 48 ) is similar to , but not identical with , the nondispersive quantity @xmath12 made use of in eq .
( 38 ) .
it is of interest to give analytic approximations for the case of low temperatures , @xmath147 .
then the quantities @xmath199 and @xmath200 are large , and it becomes convenient to make use of the approximate formula @xcite @xmath201 [ 50 ] from which it follows , in our notation , that @xmath202 [ 51 ] @xmath203 [ 52 ] @xmath204 [ 53 ] inserting eqs .
( 51)-(53 ) into eq .
( 49 ) we obtain a reasonable simple series expression for @xmath158 , at low temperatures .
a further simplification can be obtained if we keep only the leading order terms in eqs .
( 51)-(53 ) . for @xmath205
this means that we keep only the first term in each equation .
we then get in this limit @xmath206 [ 54 ] in this expression , the temperature @xmath100 is no longer present .
numerical trials show that eq .
( 54 ) is very accurate for @xmath207 .
choosing @xmath208 , we obtain from eq .
( 54 ) @xmath209 , whereas the full formula ( 49 ) yields the number -0.6530 , i. e. , an error of 0.06 per cent .
even with @xmath210 , where eq .
( 49 ) yields the number -0.678 , the error in eq .
( 54 ) is increased to only about 3 per cent .
let us summarize as follows : \(1 ) for a relativistic medium , _
i.e. _ , a medium satisfying the condition @xmath151 , the @xmath11 result for the casimir energy is most conveniently written in the form ( 8) . this expression holds for a nondispersive medium , @xmath211 being the time - splitting parameter .
the permittivity is here arbitrary ; the medium need not be dilute .
the regularization is made by subtracting off the volume terms in the two scalar green functions .
this regularization is natural , among else things because it permits one to relate @xmath19 to the surface force density @xmath10 by the relation @xmath212 .
the equivalence between the green function approach and the mode sum approach is demonstrated explicitly .
\(3 ) including the dispersive effect , the simplest dispersion relation one can imagine is the `` square '' relation given in eq .
the finite - temperature expression for the free energy on the basis of this dispersion relation is given in eq .
( 31 ) , with @xmath213 being a high frequency cutoff which is in practice dependent on the detailed structure of the medium . with reasonable accuracy @xmath137 can represent the maximum @xmath214 of the angular momentum variable @xmath125 .
typical results are shown in figs . 1 - 4 .
in the limiting case @xmath215 they are in accordance with eq .
( 33 ) , and for @xmath216 they are in accordance with eq .
in particular , the free energy is for high temperatures negative , and is a _ linear _ function of @xmath100 .
the entropy is correspondingly a ( positive ) constant .
these results are in accordance with those of refs . @xcite and @xcite .
the transition region between low - temperature and high - temperature theory can be determined with reasonable accuracy from a simple physical argument .
\(4 ) one can take spatial dispersion into account , as we have done in section 5 .
the basic dispersion relation is then eq . ( 41 ) .
a physically appealing feature of this kind of approach is that the sum over angular momenta does not have to be truncated .
typical results for this case are shown in fig .
\(6 ) a final remark is called for , as regards the numerical computation of the series in the case of spatial dispersion .
the evaluation of the expression ( 49 ) turned out to be more difficult than one might expect beforehand .
thus , a simple use of matlab turned out to be insufficient .
the problems seem to be associated with numerical noise , caused by the `` critical '' terms for which @xmath218 .
there are relatively large individual terms that almost , but not quite , compensate each other in the sum .
we managed to do the calculation making use of double - precision ms - dos quick basic . in practice ,
adequate precision was found with inclusion of less than @xmath219 terms . | the finite temperature casimir free energy is calculated for a dielectric ball of radius @xmath0 embedded in an infinite medium .
the condition @xmath1 is assumed for the inside / outside regions . both the green function method and the mode summation method are considered , and found to be equivalent . for a dilute medium
we find , assuming a simple `` square '' dispersion relation with an abrupt cutoff at imaginary frequency @xmath2 , the high temperature casimir free energy to be negative and proportional to @xmath3 . also , a physically more realistic dispersion relation involving spatial dispersion is considered , and is shown to lead to comparable results . |
stellar mass and metallicity are two of the most fundamental physical properties of galaxies .
the metal enrichment of the inter stellar medium ( ism ) of a galaxy is a consequence of supernova explosions and stellar winds and is therefore related to the star formation history ( sfh ) of the galaxy . also , the amount of mass in stars is a function of galaxy sfh .
therefore , understanding the evolution of the two properties and the relation between them is fundamental to understand the formation and evolution of galaxies .
observations have shown that for local and low redshift galaxies a tight relation exists between the galaxy mass and its metallicity ( see e.g. * ? ? ?
* ; * ? ? ?
the evolution of the mass - metallicity ( mz ) relation has been studied using emission lines from hii regions in galaxies out to @xmath2 and reveal for a galaxy of a given stellar mass a trend of a decreasing metallicity with increasing redshift @xcite . whether the numerous galaxies at the faint end of the luminosity function follow extrapolations of the mz relation can not easily be addressed using emission selected samples , and most studies have focused on composite spectra or a few individually selected massive galaxies @xcite .
gravitational lensing which magnifies the flux of background sources is a powerful tool to probe the faint end of the galaxy luminosity function .
studies of low - mass gravitationally lensed galaxies hint at a weaker evolution of the mz relation and an increasing scatter out to @xmath3 @xcite .
[ cols="<,<,^,^ , < , < , < , < " , ] + + references for measured stellar masses : @xcite for 000926 ; @xcite for 050820a ; @xcite for 070802 ; this work for 090323 . _ { e , c}$ ] ) vs velocity width for qso - dlas ( small dots ) and grb host dlas ( large blue squares )
. best fits ( using equation [ c2 ] ) are shown as dotted lines ( qso - dlas ) , full lines ( grb - dlas ) .
, title="fig:",width=328 ] -3 mm a prescription for computation of the stellar mass of qso - dla galaxies from only metallicity and redshift was given in @xcite .
@xcite , improved this prescription by adding the effect of metallicity gradients and also performed a test comparing the computed stellar mass to the measured stellar mass from the sed fits .
this test was carried out using the complete set of qso - dla galaxies for which the test is currently possible .
they concluded that the prescription is confirmed for galaxies of stellar masses down to @xmath4 , while for lower stellar masses there are no available data .
here we use the prescription from @xcite ( their equation ( 3 ) including the metallicity gradient term @xmath5 ) to compute the predicted stellar masses of all the host galaxies in our sample ( listed in table [ tab5 ] ) . for 3 of those ,
stellar masses have been determined directly via sed fitting ( also provided in table [ tab5 ] ) . for the host of grb 090323
, we use the photometric data given in @xcite and determine the stellar mass following the procedure described in @xcite and the initial mass function given in @xcite . in order to obtain the full distribution function of the allowed mass , a monte carlo simulation re - sampling the photometric errors
we measure the stellar mass for this host to be @xmath6 .
the large error bar is due to only having upper limits on the rest frame optical photometry .
we find the measured stellar masses for these four hosts to be in complete agreement with our computed values provided we use the prescription including the metallicity gradient .
if we instead use equation ( 1 ) from @xcite , which assumes a constant offset between absorption metallicity and emission metallicity , then the agreement is much poorer with the computed masses in the mean being 1 dex higher than the measured stellar masses . i.e. our data support the hypothesis that the stellar masses of grb - dla galaxies follow the same prescription as do qso - dla galaxies , and that they have metallicity gradients with a slope similar to that of qso - dlas . in a related study of a sample of 18 low redshift grb hosts with measured emission line metallicities and stellar masses from sed fits , @xcite showed that those host galaxies follow the same m - z relation as sdss galaxies , but only after correcting for the high sfr which is a result of the sfr weighted selection we discussed in section [ histos ] .
it therefore appears that the available samples of emission selected galaxies , grb selected galaxies , and dla selected galaxies follow the same m - z relations ( when corrected for their specific selection function ) and likely are drawn from the same underlying galaxy sample . in section [ histos ]
we described how the metallicity offset seen in figure [ vz - dist ] could be understood as a result of selection functions , but from figure [ zb - corrected ] it is seen that the offset could just as well be caused by the effect of metallicity gradients and different impact parameter distributions .
our sample covers a range of stellar masses from @xmath7 to @xmath8 @xmath9 , with a median of @xmath10 ( table [ tab5 ] ) .
this median mass is identical to that reported by @xcite for dla galaxies which , held together with the better fits using metallicity gradients described above , supports that at least part of the metallicity offset is a result of different impact parameter distributions . in that case the shift between the two samples seen in figure [ zb - corrected ] is most easily interpreted as the effect of different paths through the gravitational potentials .
the interpretation of the observed distribution of data points in figure [ z - corrected ] is therefore complex .
effects of redshift evolution , impact parameter distributions , metallicity gradients , and differently weighted selection functions all work to move the data - points , which causes at least part of the scatter of the relation .
we here repeat from the conclusions of @xcite that in order to move forward towards an understanding of those objects we need to identify and understand the sources of the scatter .
one of the sources ( redshift evolution ) has already been identified .
@xcite recently found that half of the scatter in their sample was removed when the effect of metallicity gradients were included .
here we have proposed that the effect of gravitational well depth could be an additional cause of scatter .
most long duration grb host galaxies display strong intrinsic dla absorption systems similar in nature to the intervening dla systems seen in qso spectra .
the grb host systems are , however , different in two ways : they originate inside the host galaxies rather than behind them and they are found at much smaller impact parameters . in addition they are also reported generally to have higher hi column densities and often to have higher metallicities than intervening dlas at the same redshift .
it is important to establish if those differences simply are a result of two different selection functions applied to the same underlying sample of high redshift galaxies , or if the two types of galaxies are truly two different populations .
we have here analysed the mass / metallicity / redshift relations of a complete literature sample of grb host galaxies and a sample of intervening dla galaxies in order to address this question .
we have found that + 1 ) the two samples are fully consistent with being drawn from the same underlying population with a single mz relation , and a single redshift evolution of this relation with a break around @xmath11 .
grb hosts are in better agreement with this redshift evolution compared to linear evolutions with constant slopes .
+ 2 ) there is evidence that the grb host galaxies have higher metallicities , but this is most likely a secondary correlation .
the primary correlation is with either impact parameter , with stellar mass , or , presumably , with a combination of the two .
the smaller impact parameters combined with a metallicity gradient will produce a metallicity offset , sfr selection bias is predicted to select galaxies of somewhat larger stellar mass than dla galaxies which will likewise cause an offset in metallicity .
+ 3 ) there is weak evidence that the @xmath12-metallicity relation for the grb hosts is offset towards larger @xmath12 values , as one would predict since their sightlines pass through a deeper part of the dark matter halo potential well than a random sightline to an intervening dla in a halo of the same mass .
it has been shown previously that qso - dlas and lyman break galaxies ( lbg ) are consistent with being drawn from the same underlying population by two very different selection functions , where qso - dlas are drawn from the very low - mass end of the lbg population @xcite . with the results presented here we have now added long duration grb hosts to this list , which means that we have made another important step towards a global description of galaxies and galaxy evolution in the early universe .
since the sample used in this pilot study is limited , it will be quite feasible to improve the accuracy of all results reported here simply by increasing the sample size .
grb host are ideally suited to shed light on the structure of high redshift galaxies .
they combine the data from emission selected galaxies directly with those of absorption selected galaxies .
i.e. we obtain in the ideal case both absorption metallicity , emission metallicity , stellar mass from sed fits , impact parameter and @xmath12 for a single galaxy .
we would like to thank jason x. prochaska for providing the hi column density for the host of grb 090313 prior to publication .
we thank an anonymous referee for a careful reading of our manuscript and for many insightful and valuable comments which significantly improved the presentation of our results .
we thank karl glazebrook and damien le borgne for helping with the stellar mass calculation of the host of grb 090323 .
ma thanks max - planck - institut fr astrophysik for hosting her during the initial part of this work .
jpuf acknowledges support from the erc - stg grant eggs-278202 .
this work was funded by an eso dgdf grant to pm and wf .
the dark cosmology centre is funded by the danish national research foundation .
asplund m. , grevesse n. , sauval a.j . , scott p. , 2009
, ara&a , 47 , 481 baldry i.k . , glazebrok k. , 2003 , apj , 593 , 258 belli s. , jones t. , ellis r.s . , richard j. , 2013 , apj , 772 , 141 berger e. , penprase b.e . , cenko s.b . ,
kulkarni s.r . ,
, steidel c.c . ,
reddy n.a . , 2006 , apj , 642 , 979 bird s. , haehnelt m. , neeleman m. , genel s. , vogelsberger m. , hernquist l. , 2014 , arxiv1407.7858b bloom j.s . ,
kulkarni s.r . , djorgovski s.g . , 2002 ,
apj , 123 , 1111 castro s. , galama t.j . , harrison f.a . , holtzman j.a . , bloom j.s . ,
djorgovski s.g . ,
kulkarni s.r .
, 2003 , apj , 586 , 128 chen h. et al . , 2009 ,
apj , 691 , 152 christensen l. , mller p. , fynbo j.p.u . , zafar t. , 2014 , arxiv1404.6529 christensen et al . , 2012 , mnras , 427 , 1973 christensen l. , hjorth j. , gorosabel j.,2004 , a&a , 425 , 913 chornock r. , berger e. , fox d.b . , fong w. , laskar t. , roth k.c . , 2014 , arxiv1405.7400 cucchiara a. , fumagalli m. , rafelski m. , kocevski d. , prochaska j.x . , cooke r.j . , becker g.d . , 2014 , arxiv1408.3578 cullen f. , cirasuolo m. , mclure r.j . , dunlop j.s . , bowler r.a.a
, 2014 , mnras , 440 , 2300 de cia a. et al . , 2011 , mnras , 412 , 2229 delia v. et al . , 2014 , a&a , 564 , 38 delia v. , campana s. , covino s. , davanzo p. , piranomonte s. , tagliaferri g. , 2011 , mnras , 418 , 680 delia v. et al . , 2010 ,
a&a , 523 , 36 dekel a. , woo j. , 2003 , mnras , 344 , 1131 de ugarte postigo a. et el .
, 2010 , a&a , 513 , 42 eladdttir a. et al . , 2009 , apj , 697 , 1725 ellison s. l. , kanekar n. , prochaska j.x .
, momjian e. , worseck g. , 2012 , mnras , 424 , 293 erb d.k . ,
shapley a.e . , pettini m. , steide , c.c . ,
reddy n.a . , adelberger k.l . , 2006 ,
apj , 644 , 813 fox a.j .
, ledoux c. , vreeswijk p.m. , smette a. , jaunsen a.o . , 2008 ,
a&a , 491 , 189 fruchter a.s .
et al . , 2006 ,
natur , 441 , 463 fynbo , j.p.u .
et al . , 2013 ,
mnras , 436 , 361 fynbo j.p.u .
et al . , 2009 , apjss , 185 , 526 fynbo j.p.u . , prochaska j.x .
, sommer - larsen j. , dessauges - zavadsky m. , mller p. , 2008 ,
apj , 683,321 fynbo j.p.u .
et al . , 2006 ,
a&a , 451 , 47 fynbo j.p.u .
, mller p. , warren s.j .
, 1999 , mnras , 305 , 849 glazebrook k. et al . , 2004 , natur , 430 , 181 graham j.f . , fruchter a.s .
, 2013 , apj , 774 , 119 henry a. et al .
, 2013 , apj , 776 , 27h hjorth j. et al . , 2012 , apj , 756 , 187 krogager j .- k .
et al . , 2013 , mnras , 433 , 3091 krogager j .- k . , fynbo j.p.u . , mller p. , ledoux c. , noterdaeme p. , christensen l. , milvang - jensen b. , sparre m. , 2012 , mnras , 424 , 1 krhler et al . , 2013 ,
a&a , 557 , 18 krhler et al . , 2011 ,
a&a , 534 , 108 ledoux c. , vreeswijk p.m. , smette a. , fox a.j . ,
petitjean p. , ellison s.l .
, fynbo j.p.u . , savaglio s. , 2009 , a&a , 506 , 661 ledoux c. , petitjean p. , fynbo j.p.u . , mller p. , srianand r. , 2006 , a&a , 457 , 71 ledoux c. , bergeron j. , petitjean p. , 2002 ,
a&a , 385 , 802 le floch e. et al . , 2003 , a&a , 400 , 499 maiolino r. , et al . , 2008 , a&a , 488 , 463 mannucci f. , salvaterra r. , campisi m.a . , 2011 , mnras , 414 , 1263 michaleowskiowski m.j .
et al . , 2012 ,
apj , 755 , 85 mcbreen s. et al . , 2010 , a&a , 516 , 71 mller p. , fynbo j.p.u .
, ledoux c. , nilsson k.k . , 2013 , mnras , 430 , 2680 mller p. , warren s.j . ,
fall s.m . ,
fynbo j.p.u . , jakobsen p. , 2002 ,
apj , 574 , 51 mller p. , warren s.j . , 1998 , mnras , 299 , 661 neeleman m. , wolfe a.m. , prochaska j.x . , rafelski m. , 2013 , apj , 769 , 54 perley d.a .
et al . , 2013 , apj , 778 , 128 proux c. , meiring j.d . , kulkarni v.p . , khare p. , lauroesch j.t . , vladilo g. , york d.g . , 2008 , mnras , 386 , 2209 proux c. , meiring j.d .
, kulkarni v.p . , ferlet r. , khare p. , lauroesch j.t . , vladilo g. , york , d.g . , 2006 , mnras , 372 , 369 pettini m. , ellison s.l .
, steidel c.c .
, shapley a.e . ,
bowen d.v .
, 2000 , apj , 532 , 65 pettini m. , smith l.j .
, king d.l . , hunstead r.w . , 1997 , apj , 486 , 665 piranomonte s. et al . , 2008 ,
a&a , 492,775 pontzen a. et al .
2008 , mnras , 390 , 1349 price p.a .
et al . , 2007 ,
apj , 663 , 57 prochaska j.x .
, chen h. , wolfe a.m. , dessauges - zavadsky m. , bloom j.s . , 2008 , apj , 672 , 59 prochaska j.x . , chen h. , dessauges - zavadsky m. , bloom j.s . , 2007a , apj , 666 , 267 prochaska j.x .
et al . , 2007b , apjs , 168 , 231 prochaska j.x . , gawiser e. , wolfe a.m. , castro s. , djorgovski s.g . , 2003 ,
apj , 595 , 9 prochaska j.x . , wolfe a.m. , 1997 , apj , 487 , 73 rafelski m. , wolfe a.m. , prochaska j.x . , neeleman m. , mendez a.j . , 2012 ,
apj , 755 , 89 richard j. , jones t. , ellis r. , stark d.p . , livermore r. , swinbank m. , 2011 , mnras , 413 , 643 savaglio s. et al . , 2012 , mnras , 420 , 627 savaglio s. , glazebrook k. , le borgne d. , 2009 , apj , 691 , 182 savaglio s. , 2006 , njp , 8 , 195 savaglio s. et al . , 2005 , apj , 635 , 260 savaglio s. , fall s.m . , fiore f. , 2003 , apj , 585 , 638 sparre et al . , 2014 , apj , 785 , 150 swinbank a.m. , sobral d. , smail ian , geach j.e .
, best p.n .
, mccarthy i.g . , crain r.a . , theuns t. , 2012 , mnras , 426 , 935 tescari e. , viel m. , tornatore l. , borgani s. , 2009 , mnras , 397 , 411 thne c.c .
et al . , 2013 ,
mnras , 428 , 3590 tremonti et al .
2004 , apj , 613 , 898 troncoso p. , et al . , 2014 ,
a&a , 563 , 58 van zee l. , salzer j.j . , haynes m.p . ,
odonoghue a.a .
, balonek t.j . , 1998 ,
aj , 116 , 2805 vreeswijk p.m. et al .
, 2004 , a&a , 419 , 927 watson d. et al .
, 2006 , apj , 652 , 1011 wuyts e. , rigby j.r . , sharon k. , gladders m.d . , 2012 ,
apj , 755 , 73 zwaan m.a .
, van der hulst j.m . ,
briggs f.h . ,
verheijen m.a.w .
, ryan - weber e.v . , 2005 , mnras , 364 , 1467 | we analyze a sample of 16 absorption systems intrinsic to long duration grb host galaxies at @xmath0 for which the metallicities are known .
we compare the relation between the metallicity and cold gas velocity width for this sample to that of the qso - dlas , and find complete agreement .
we then compare the redshift evolution of the mass - metallicity relation of our sample to that of qso - dlas and find that also grb hosts favour a late onset of this evolution , around a redshift of @xmath1 .
we compute predicted stellar masses for the grb host galaxies using the prescription determined from qso - dla samples and compare the measured stellar masses for the four hosts where stellar masses have been determined from sed fits .
we find excellent agreement and conclude that , on basis of all available data and tests , long duration grb - dla hosts and intervening qso - dlas are consistent with being drawn from the same underlying population .
grb host galaxies and qso - dlas are found to have different impact parameter distributions and we briefly discuss how this may affect statistical samples .
the impact parameter distribution has two effects .
first any metallicity gradient will shift the measured metallicity away from the metallicity in the centre of the galaxy , second the path of the sightline through different parts of the potential well of the dark matter halo will cause different velocity fields to be sampled .
we report evidence suggesting that this second effect may have been detected .
[ firstpage ] galaxies : high - redshift galaxies : ism galaxies : star formation galaxies :
evolution galaxies : formation quasars : absorption lines |
all musical pieces are similar , but some are more similar than others . apart from being an infinite source of discussion ( `` haydn is just like mozart no , he s not ! '' ) , such similarities are also crucial for the design of efficient music information retrieval systems .
the amount of digitized music available on the internet has grown dramatically in recent years , both in the public domain and on commercial sites .
napster and its clones are prime examples .
websites offering musical content in some form or other ( mp3 , midi , ) need a way to organize their wealth of material ; they need to somehow classify their files according to musical genres and subgenres , putting similar pieces together .
the purpose of such organization is to enable users to navigate to pieces of music they already know and like , but also to give them advice and recommendations ( `` if you like this , you might also like '' ) .
currently , such organization is mostly done manually by humans , but some recent research has been looking into the possibilities of automating music classification .
a human expert , comparing different pieces of music with the aim to cluster likes together , will generally look for certain specific similarities .
previous attempts to automate this process do the same . generally speaking
, they take a file containing a piece of music and extract from it various specific numerical features , related to pitch , rhythm , harmony etc .
one can extract such features using for instance fourier transforms @xcite or wavelet transforms @xcite .
the feature vectors corresponding to the various files are then classified or clustered using existing classification software , based on various standard statistical pattern recognition classifiers @xcite , bayesian classifiers @xcite , hidden markov models @xcite , ensembles of nearest - neighbor classifiers @xcite or neural networks @xcite .
for example , one feature would be to look for rhythm in the sense of beats per minute .
one can make a histogram where each histogram bin corresponds to a particular tempo in beats - per - minute and the associated peak shows how frequent and strong that particular periodicity was over the entire piece . in @xcite we see a gradual change from a few high peaks to many low and spread - out ones going from hip - hip , rock , jazz , to classical .
one can use this similarity type to try to cluster pieces in these categories .
however , such a method requires specific and detailed knowledge of the problem area , since one needs to know what features to look for .
our aim is much more general .
we do not look for similarity in specific features known to be relevant for classifying music ; instead we apply a general mathematical theory of similarity .
the aim is to capture , in a single similarity metric , _ every effective metric _ : effective versions of hamming distance , euclidean distance , edit distances , lempel - ziv distance , and so on .
such a metric would be able to simultaneously detect _ all _ similarities between pieces that other effective metrics can detect .
rather surprisingly , such a `` universal '' metric indeed exists .
it was developed in @xcite , based on the `` information distance '' of @xcite .
roughly speaking , two objects are deemed close if we can significantly `` compress '' one given the information in the other , the idea being that if two pieces are more similar , then we can more succinctly describe one given the other . here compression is based on the ideal mathematical notion of kolmogorov complexity , which unfortunately is not effectively computable .
it is well known that when a pure mathematical theory is applied to the real world , for example in hydrodynamics or in physics in general , we can in applications only approximate the theoretical ideal .
but still the theory gives a framework and foundation for the applied science .
similarly here .
we replace the ideal but noncomputable kolmogorov - based version by standard compression techniques .
we lose theoretical optimality in some cases , but gain an efficiently computable similarity metric intended to approximate the theoretical ideal .
in contrast , a later and partially independent compression - based approach of @xcite for building language - trees while citing @xcite is by _ ad hoc _ arguments about empirical shannon entropy and kullback - leibler distance resulting in non - metric distances . earlier research has demonstrated that this new universal similarity metric works well on concrete examples in very different application fields the first completely automatic construction of the phylogeny tree based on whole mitochondrial genomes , @xcite and a completely automatic construction of a language tree for over 50 euro - asian languages @xcite .
other applications , not reported in print , are detecting plagiarism in student programming assignments @xcite , and phylogeny of chain letters .
in this paper we apply this compression - based method to the classification of pieces of music .
we perform various experiments on sets of mostly classical pieces given as midi ( musical instrument digital interface ) files .
this contrasts with most earlier research , where the music was digitized in some wave format or other ( the only other research based on midi that we are aware of is @xcite ) .
we compute the distances between all pairs of pieces , and then build a tree containing those pieces in a way that is consistent with those distances . first ,
as proof of principle , we run the program on three artificially generated data sets , where we know what the final answer should be .
the program indeed classifies these perfectly .
secondly , we show that our program can distinguish between various musical genres ( classical , jazz , rock ) quite well .
thirdly , we experiment with various sets of classical pieces .
the results are quite good ( in the sense of conforming to our expectations ) for small sets of data , but tend to get a bit worse for large sets .
considering the fact that the method knows nothing about music , or , indeed , about any of the other areas we have applied it to elsewhere , one is reminded of dr johnson s remark about a dog s walking on his hind legs : `` it is not done well ; but you are surprised to find it done at all . ''
the paper is organized as follows .
we first give a domain - independent overview of compression - based clustering : the ideal distance metric based on kolmogorov complexity , and the quartet method that turns the matrix of distances into a tree . in section [ secdetails ]
we give the details of the current application to music , the specific file formats used etc . in section [ secresults ]
we report the results of our experiments .
we end with some directions for future research .
each object ( in the application of this paper : each piece of music ) is coded as a string @xmath0 over a finite alphabet , say the binary alphabet .
the integer @xmath1 gives the length of the shortest compressed binary version from which @xmath0 can be fully reproduced , also known as the _ kolmogorov complexity _ of @xmath0 .
`` shortest '' means the minimum taken over every possible decompression program , the ones that are currently known as well as the ones that are possible but currently unknown .
we explicitly write only `` decompression '' because we do not even require that there is also a program that compresses the original file to this compressed version
if there is such a program then so much the better . technically , the definition of kolmogorov complexity is as follows .
first , we fix a syntax for expressing all and only computations ( computable functions ) .
this can be in the form of an enumeration of all turing machines , but also an enumeration of all syntactically correct programs in some universal programming language like java , lisp , or c. we then define the kolmogorov complexity of a finite binary string as the length of the shortest turing machine , java program , etc . in our chosen syntax .
which syntax we take is unimportant , but we have to stick to our choice .
this choice attaches a definite positive integer as the kolmogorov complexity to each finite string . though defined in terms of a particular machine model ,
the kolmogorov complexity is machine - independent up to an additive constant and acquires an asymptotically universal and absolute character through church s thesis , and from the ability of universal machines to simulate one another and execute any effective process .
the kolmogorov complexity of an object can be viewed as an absolute and objective quantification of the amount of information in it .
this leads to a theory of _ absolute _ information _ contents _ of _ individual _ objects in contrast to classic information theory which deals with _ average _ information _ to communicate _ objects produced by a _
random source_. so @xmath1 gives the length of the ultimate compressed version , say @xmath2 , of @xmath0 .
this can be considered as the amount of information , number of bits , contained in the string .
similarly , @xmath3 is the minimal number of bits ( which we may think of as constituting a computer program ) required to reconstruct @xmath0 from @xmath4 . in a way
@xmath1 expresses the individual `` entropy '' of @xmath0the minimal number of bits to communicate @xmath0 when sender and receiver have no knowledge where @xmath0 comes from .
for example , to communicate mozart s `` zauberflte '' from a library of a million items requires at most 20 bits ( @xmath5 ) , but to communicate it from scratch requires megabits . for more details on this pristine notion of individual information
content we refer to the textbook @xcite .
as mentioned , our approach is based on a new very general similarity distance , classifying the objects in clusters of objects that are close together according to this distance . in mathematics ,
lots of different distances arise in all sorts of contexts , and one usually requires these to be a ` metric ' , since otherwise undesirable effects may occur .
a metric is a distance function @xmath6 that assigns a non - negative distance @xmath7 to any two objects @xmath8 and @xmath9 , in such a way that 1 .
@xmath10 only where @xmath11 2 .
@xmath12 ( symmetry ) 3 .
@xmath13 ( triangle inequality ) a familiar example of a metric is the euclidean metric , the everyday distance @xmath14 between two objects @xmath15 expressed in , say , meters .
clearly , this distance satisfies the properties @xmath16 , @xmath17 , and @xmath18 ( substitute @xmath19 amsterdam , @xmath20 brussels , and @xmath21 chicago . )
we are interested in `` similarity metrics '' .
for example , if the objects are classical music pieces then the function @xmath10 if @xmath8 and @xmath9 are by the same composer and @xmath22 otherwise , is a similarity metric , albeit a somewhat elusive one .
this captures only one , but quite a significant , similarity aspect between music pieces . in @xcite , a new theoretical approach to a wide class of similarity metrics
was proposed : the `` normalized information distance '' is a metric , and it is universal in the sense that this single metric uncovers all similarities simultaneously that the metrics in the class uncover separately .
this should be understood in the sense that if two pieces of music are similar ( that is , close ) according to the particular feature described by a particular metric , then they are also similar ( that is , close ) in the sense of the normalized information distance metric .
this justifies calling the latter _ the _ similarity metric .
oblivious to the problem area concerned , simply using the distances according to the similarity metric , our method fully automatically classifies the objects concerned , be they music pieces , text corpora , or genomic data .
more precisely , the approach is as follows .
each pair of such strings @xmath0 and @xmath4 is assigned a distance @xmath23 there is a natural interpretation to @xmath24 : if , say , @xmath25 then we can rewrite @xmath26 where @xmath27 is the information in @xmath4 about @xmath0 satisfying the symmetry property @xmath28 up to a logarithmic additive error @xcite .
that is , the distance @xmath24 between @xmath0 and @xmath4 is the number of bits of information that is not shared between the two strings per bit of information that could be maximally shared between the two strings .
it is clear that @xmath24 is symmetric , and in @xcite it is shown that it is indeed a metric .
moreover , it is universal in the sense that every metric expressing some similarity that can be computed from the objects concerned is comprised ( in the sense of minorized ) by @xmath24 .
it is these distances that we will use , albeit in the form of a rough approximation : for @xmath1 we simply use standard compression software like ` gzip ' , ` bzip2 ' , or ` compress ' . to compute the conditional version , @xmath3 we use a sophisticated theorem , known as `` symmetry of algorithmic information '' in @xcite
this says @xmath29 so to compute the conditional complexity @xmath3 we can just take the difference of the unconditional complexities @xmath30 and @xmath31 .
this allows us to approximate @xmath24 for every pair @xmath32 .
our actual practice falls short of the ideal theory in at least three respects : \(i ) the claimed universality of the similarity distance @xmath24 holds only for indefinitely long sequences @xmath32 .
once we consider strings @xmath32 of definite length @xmath33 , the similarity distance is only universal with respect to `` simple '' computable normalized information distances , where `` simple '' means that they are computable by programs of length , say , logarithmic or polylogarithmic in @xmath33 .
this reflects the fact that , technically speaking , the universality is achieved by summing the weighted contribution of all similarity distances in the class considered with respect to the objects considered .
only similarity distances of which the complexity is small ( which means that the weight is large ) with respect to the size of the data concerned kick in .
\(ii ) the kolmogorov complexity is not computable , and it is in principle impossible to compute how far off our approximation is from the target value in any useful sense .
\(iii ) to approximate the information distance in a practical sense we use the standard compression program bzip2 .
while better compression of a string will always approximate the kolmogorov complexity better , this is , regrettably , not true for the ( normalized ) information distance .
namely , using ( [ eq.condition ] ) we consider the difference of two compressed quantities .
different compressors may compress the two quantities differently , causing an increase in the difference even when both quantities are compressed better ( but not both as well ) . in the normalized information distance
we also have to deal with a ratio that causes the same problem .
thus , a better compression program may not necessarily mean that we also approximate the ( normalized ) information distance better .
this was borne out by the results of our experiments using different compressors . despite these caveats it turns out that the practice inspired by the rigorous ideal theory performs quite well .
we feel this is an example that an _ ad hoc _ approximation guided by a good theory is preferable above _ ad hoc _ approaches without underlying theoretical foundation .
the above approach allows us to compute the distance between any pair of objects ( any two pieces of music ) .
we now need to cluster the objects , so that objects that are similar according to our metric are placed close together .
we do this by computing a phylogeny tree based on these distances . such a phylogeny tree can represent evolution of species but more widely simply accounts for closeness of objects from a set with a distance metric . such a tree will group objects in subtrees : the clusters . to find the phylogeny tree
there are many methods .
one of the most popular is the quartet method .
the idea is as follows : we consider every group of four elements from our set of @xmath33 elements ( in this case , musical pieces ) ; there are @xmath34 such groups . from each group
@xmath35 we construct a tree of arity 3 , which implies that the tree consists of two subtrees of two leaves each .
let us call such a tree a _
quartet_. there are three possibilities denoted ( i ) @xmath36 , ( ii ) @xmath37 , and ( iii ) @xmath38 , where a vertical bar divides the two pairs of leaf nodes into two disjoint subtrees ( figure [ figquart ] ) .
the cost of a quartet is defined as the sum of the distances between each pair of neighbors ; that is , @xmath39 .
for any given tree @xmath40 and any group of four leaf labels @xmath35 , we say @xmath40 is @xmath41 with @xmath36 if and only if the path from @xmath42 to @xmath43 does not cross the path from @xmath44 to @xmath0 .
note that exactly one of the three possible quartets for any set of 4 labels must be consistent for any given tree .
we may think of a large tree having many smaller quartet trees embedded within its structure ( figure [ figquartex ] ) .
the total cost of a large tree is defined to be the sum of the costs of all consistent quartets .
first , generate a list of all possible quartets for all groups of labels under consideration .
for each group of three possible quartets for a given set of four labels , calculate a best ( minimal ) cost , and a worst ( maximal ) cost .
summing all best quartets yields the best ( minimal ) cost .
conversely , summing all worst quartets yields the worst ( maximal ) cost .
the minimal and maximal values need not be attained by actual trees , however the score of any tree will lie between these two values . in order to be able to compare tree scores in a more uniform way , we now rescale the score linearly such that the worst score maps to 0 , and the best score maps to 1 , and term this the _ normalized tree benefit score _ @xmath45 . the goal of the quartet method is to find a full tree with a maximum value of @xmath45 , which is to say , the lowest total cost .
this optimization problem is known to be np - hard @xcite ( which means that it is infeasible in practice ) but we can sometimes solve it , and always approximate it .
the current methods in @xcite are far too computationally intensive ; they run many months or years on moderate - sized problems of 30 objects .
we have designed a simple method based on randomization and hill - climbing .
first , a random tree with @xmath46 nodes is created , consisting of @xmath33 leaf nodes ( with 1 connecting edge ) labeled with the names of musical pieces , and @xmath47 non - leaf or _
internal _ nodes labeled with the lowercase letter `` n '' followed by a unique integer identifier .
each internal node has exactly three connecting edges . for this tree @xmath40
, we calculate the total cost of all consistent quartets , and invert and scale this value to find @xmath45 .
typically , a random tree will be consistent with around @xmath48 of all quartets .
now , this tree is denoted the currently best known tree , and is used as the basis for further searching .
we define a simple mutation on a tree as one of the three possible transformations : 1 . a _ leaf swap _ , which consists of randomly choosing two leaf nodes and swapping them .
2 . a _ subtree swap _ , which consists of randomly choosing two internal nodes and swapping the subtrees rooted at those nodes .
3 . a _ subtree transfer _ , whereby a randomly chosen subtree ( possibly a leaf ) is detached and reattached in another place , maintaining arity invariants .
each of these simple mutations keeps invariant the number of leaf and internal nodes in the tree ; only the structure and placements change .
define a full mutation as a sequence of at least one but potentially many simple mutations , picked according to the following distribution .
first we pick the number @xmath49 of simple mutations that we will perform with probability @xmath50 . for each such simple mutation , we choose one of the three types listed above with equal probability .
finally , for each of these simple mutations , we pick leaves or internal nodes , as necessary .
notice that trees which are close to the original tree ( in terms of number of simple mutation steps in between ) are examined often , while trees that are far away from the original tree will eventually be examined , but not very frequently .
so in order to search for a better tree , we simply apply a full mutation on @xmath40 to arrive at @xmath51 , and then calculate @xmath52 .
if @xmath53 , then keep @xmath51 as the new best tree . otherwise , try a new different tree and repeat .
if @xmath52 ever reaches @xmath54 , then halt , outputting the best tree .
otherwise , run until it seems no better trees are being found in a reasonable amount of time , in which case the approximation is complete .
note that if a tree is ever found such that @xmath55 , then we can stop because we can be certain that this tree is optimal , as no tree could have a lower cost .
in fact , this perfect tree result is achieved in our artificial tree reconstruction experiment ( section [ sect.artificial ] ) reliably in less than ten minutes . for real - world data
, @xmath45 reaches a maximum somewhat less than @xmath54 , presumably reflecting inconsistency in the distance matrix data fed as input to the algorithm , or indicating a search space too large to solve exactly . on many typical problems of up to 40 objects
this tree - search gives a tree with @xmath56 within half an hour . for large numbers of objects , tree scoring itself
can be slow ( as this takes order @xmath57 computation steps ) , and the space of trees is also large , so the algorithm may slow down substantially . for larger experiments
, we use a c++/ruby implementation with mpi ( message passing interface , a common standard used on massively parallel computers ) on a cluster of workstations in parallel to find trees more rapidly .
we can consider the graph of figure [ figprogress ] , mapping the achieved @xmath45 score as a function of the number of trees examined .
progress occurs typically in a sigmoidal fashion towards a maximal value @xmath58 .
a problem with the outcomes is as follows : for natural data sets we often see some leaf nodes ( data items ) placed near the center of the tree as singleton leaves attached to internal nodes , without sibling leaf nodes .
this results in a more linear , stretched out , and less balanced , tree .
such trees , even if they represent the underlying distance matrix faithfully , are hard to fully understand and may cause misunderstanding of represented relations and clusters . to counteract this effect , and to bring out the clusters of related items more visibly , we have added a penalty term of the following form : for each internal node with exactly one leaf node attached , the tree s score is reduced by 0.005 .
this induces a tendency in the algorithm to avoid producing degenerate mostly - linear trees in the face of data that is somewhat inconsistent , and creates balanced and more illuminating clusters .
it should be noted that the penalty term causes the algorithm in some cases to settle for a slightly lower @xmath45 score than it would have without penalty term .
also the value of the penalty term is heuristically chosen .
the largest experiment used 60 items , and we typically had only a couple of orphans causing a penalty of only a few percent .
this should be set off against the final @xmath45 score of above 0.85 .
another practicality concerns the stopping criterion , at which @xmath45 value we stop .
essentially we stopped when the @xmath45 value did nt change after examining a large number of mutated trees .
an example is the progress of figure [ figprogress ] ,
initially , we downloaded 118 separate midi ( musical instrument digital interface , a versatile digital music format available on the world - wide - web ) files selected from a range of classical composers , as well as some popular music .
each of these files was run through a preprocessor to extract just midi note - on and note - off events .
these events were then converted to a player - piano style representation , with time quantized in @xmath59 second intervals .
all instrument indicators , midi control signals , and tempo variations were ignored . for each track in the midi file , we calculate two quantities : an _ average volume _ and a _ modal note_.
the average volume is calculated by averaging the volume ( midi note velocity ) of all notes in the track .
the modal note is defined to be the note pitch that sounds most often in that track .
if this is not unique , then the lowest such note is chosen .
the modal note is used as a key - invariant reference point from which to represent all notes .
it is denoted by @xmath60 , higher notes are denoted by positive numbers , and lower notes are denoted by negative numbers .
a value of @xmath54 indicates a half - step above the modal note , and a value of @xmath61 indicates a whole - step below the modal note .
the tracks are sorted according to decreasing average volume , and then output in succession . for each track , we iterate through each time sample in order , outputting a single signed 8-bit value for each currently sounding note .
two special values are reserved to represent the end of a time step and the end of a track .
this file is then used as input to the compression stage for distance matrix calculation and subsequent tree search .
with the natural data sets of music pieces that we use , one may have the preconception ( or prejudice ) that music by bach should be clustered together , music by chopin should be clustered together , and so should music by rock stars .
however , the preprocessed music files of a piece by bach and a piece by chopin , or the beatles , may resemble one another more than two different pieces by bach by accident or indeed by design and copying .
thus , natural data sets may have ambiguous , conflicting , or counterintuitive outcomes . in other words ,
the experiments on actual pieces have the drawback of not having one clear `` correct '' answer that can function as a benchmark for assessing our experimental outcomes . before describing the experiments we did with midi files of actual music , we discuss three experiments that show that our program indeed does what it is supposed to do at least in artificial situations where we know in advance what the correct answer is .
the similarity machine consists of two parts : ( i ) extracting a distance matrix from the data , and ( ii ) constructing a tree from the distance matrix using our novel quartet - based heuristic .
* testing the quartet - based tree construction : * we first test whether the quartet - based tree construction heuristic is trustworthy : we generated a random ternary tree @xmath40 with 18 leaves , and derived a distance metric from it by defining the distance between two nodes as follows : given the length of the path from @xmath8 to @xmath9 , in an integer number of edges , as @xmath62 , let @xmath63 except when @xmath64 , in which case @xmath65 .
it is easy to verify that this simple formula always gives a number between 0 and 1 , and is monotonic with path length .
given only the @xmath66 matrix of these normalized distances , our quartet method exactly reconstructed @xmath40 represented in figure [ figarttreereal ] , with @xmath67 . * testing the similarity machine on artificial data : * given that the tree reconstruction method is accurate on clean consistent data , we tried whether the full procedure works in an acceptable manner when we know what the outcome should be like : we randomly generated 22 separate 1-kilobyte blocks of data where each byte was equally probable and called these _ tags_. each tag was associated with a different lowercase letter of the alphabet .
next , we generated 80-kilobyte files by starting with a block of purely random bytes and applying one , two , three , or four different tags on it . applying a tag consists of ten repetitions of picking a random location in the 80-kilobyte file , and overwriting that location with the universally consistent tag that is indicated .
so , for instance , to create the file referred to in the diagram by `` a '' , we start with 80 kilobytes of random data , then pick ten places to copy over this random data with the arbitrary 1-kilobyte sequence identified as tag _
a_. similarly , to create file `` ab '' , we start with 80 kilobytes of random data , then pick ten places to put copies of tag _ a _ , then pick ten more places to put copies of tag _ b _ ( perhaps overwriting some of the _ a _ tags ) . because we never use more than four different tags , and therefore never place more than 40 copies of tags , we can expect that at least half of the data in each file is random and uncorrelated with the rest of the files .
the rest of the file is correlated with other files that also contain tags in common ; the more tags in common , the more related the files are .
the resulting tree is given in figure [ figtaggedfiles ] ; it can be seen that clustering occurs exactly as we would expect .
the @xmath45 score is 0.905 .
* testing the similarity machine on natural data : * we test gross classification of files based on markedly different file types . here
, we chose several files : 1 .
four mitochondrial gene sequences , from a black bear , polar bear , fox , and rat .
four excerpts from the novel _ the zeppelin s passenger _ by e. phillips oppenheim 3 .
four midi files without further processing ; two from jimi hendrix and two movements from debussy s suite bergamasque 4 . two linux x86 elf executables ( the _ cp _ and _ rm _ commands ) 5 . two compiled java class files .
as expected , the program correctly classifies each of the different types of files together with like near like .
the result is reported in figure [ figfiletypes ] with @xmath45 equal to 0.984 . before testing whether our program can see the distinctions between various classical composers ,
we first show that it can distinguish between three broader musical genres : classical music , rock , and jazz .
this should be easier than making distinctions `` within '' classical music .
all musical pieces we used are listed in the tables in the appendix . for the genre - experiment we used 12 classical pieces ( the small set from table [ tableclassicalpieces ] , consisting of bach , chopin , and debussy ) , 12 jazz pieces ( table [ tablejazzpieces ] ) , and 12 rock pieces ( table [ tablerockpieces ] ) .
the tree that our program came up with is given in figure [ figgenres ] .
the @xmath45 score is 0.858 .
the discrimination between the 3 genres is good but not perfect .
the upper branch of the tree contains 10 of the 12 jazz pieces , but also chopin s prlude no . 15 and a bach prelude .
the two other jazz pieces , miles davis `` so what '' and john coltrane s `` giant steps '' are placed elsewhere in the tree , perhaps according to some kinship that now escapes us but can be identified by closer studying of the objects concerned . of the rock pieces ,
9 are placed close together in the rightmost branch , while hendrix s `` voodoo chile '' , rush `` yyz '' , and dire straits `` money for nothing '' are further away . in the case of the hendrix piece
this may be explained by the fact that it does not fit well in a specific genre .
most of the classical pieces are in the lower left part of the tree .
surprisingly , 2 of the 4 bach pieces are placed elsewhere .
it is not clear why this happens and may be considered an error of our program , since we perceive the 4 bach pieces to be very close , both structurally and melodically ( as they all come from the mono - thematic `` wohltemperierte klavier '' ) .
however , bach s is a seminal music and has been copied and cannibalized in all kinds of recognizable or hidden manners ; closer scrutiny could reveal likenesses in its present company that are not now apparent to us .
in effect our similarity engine aims at the ideal of a perfect data mining process , discovering unknown features in which the data can be similar . in table
[ tableclassicalpieces ] we list all 60 classical piano pieces used , together with their abbreviations .
some of these are complete compositions , others are individual movements from larger compositions .
they all are piano pieces , but experiments on 34 movements of symphonies gave very similar results ( section [ secsymphonies ] ) .
apart from running our program on the whole set of 60 piano pieces , we also tried it on two smaller sets : a small 12-piece set , indicated by ` ( s ) ' in the table , and a medium - size 32-piece set , indicated by ` ( s ) ' or ` ( m ) ' .
the small set encompasses the 4 movements from debussy s suite bergamasque , 4 movements of book 2 of bach s wohltemperierte klavier , and 4 preludes from chopin s opus 28 .
as one can see in figure [ figsmallset ] , our program does a pretty good job at clustering these pieces .
the @xmath45 score is also high : 0.958 .
the 4 debussy movements form one cluster , as do the 4 bach pieces .
the only imperfection in the tree , judged by what one would intuitively expect , is that chopin s prlude no . 15 lies a bit closer to bach than to the other 3 chopin pieces .
this prlude no 15 , in fact , consistently forms an odd - one - out in our other experiments as well .
this is an example of pure data mining , since there is some musical truth to this , as no .
15 is perceived as by far the most eccentric among the 24 prludes of chopin s opus 28 . the medium set adds 20 pieces to the small set : 6 additional bach pieces , 6 additional chopins , 1 more debussy piece , and 7 pieces by haydn .
the experimental results are given in figure [ figmediumset ] .
the @xmath45 score is slightly lower than in the small set experiment : 0.895 .
again , there is a lot of structure and expected clustering .
most of the bach pieces are together , as are the four debussy pieces from the suite bergamasque .
these four should be together because they are movements from the same piece ; the fifth debussy item is somewhat apart since it comes from another piece .
both the haydn and the chopin pieces are clustered in little sub - clusters of two or three pieces , but those sub - clusters are scattered throughout the tree instead of being close together in a larger cluster .
these small clusters may be an imperfection of the method , or , alternatively point at musical similarities between the clustered pieces that transcend the similarities induced by the same composer .
indeed , this may point the way for further musicological investigation .
figure [ figlargeset ] gives the output of a run of our program on the full set of 60 pieces .
this adds 10 pieces by beethoven , 8 by buxtehude , and 10 by mozart to the medium set .
the experimental results are given in figure [ figlargeset ] .
the results are still far from random , but leave more to be desired than the smaller - scale experiments . indeed , the @xmath45 score has dropped further from that of the medium - sized set to 0.844 .
this may be an artifact of the interplay between the relatively small size , and large number , of the files compared : ( i ) the distances estimated are less accurate ; ( ii ) the number of quartets with conflicting requirements increases ; and ( iii ) the computation time rises to such an extent that the correctness score of the displayed cluster graph within the set time limit is lower than in the smaller samples .
nonetheless , bach and debussy are still reasonably well clustered , but other pieces ( notably the beethoven and chopin ones ) are scattered throughout the tree .
maybe this means that individual music pieces by these composers are more similar to pieces of other composers than they are to each other ?
the placement of the pieces is closer to intuition on a small level ( for example , most pairing of siblings corresponds to musical similarity in the sense of the same composer ) than on the larger level .
this is similar to the phenomenon of little sub - clusters of haydn or chopin pieces that we saw in the medium - size experiment . finally , we tested whether the method worked for more complicated music , namely 34 symphonic pieces .
we took two haydn symphonies ( no .
95 in one file , and the four movements of 104 ) , three mozart symphonies ( 39 , 40 , 41 ) , three beethoven symphonies ( 3 , 4 , 5 ) , of schubert s unfinished symphony , and of saint - saens symphony no .
the results are reported in figure [ figsymphonies ] , with a quite reasonable @xmath45 score of 0.860 .
our research raises many questions worth looking into further : * the program can be used as a data mining machine to discover hitherto unknown similarities between music pieces of different composers or indeed different genres . in this manner
we can discover plagiarism or indeed honest influences between music pieces and composers .
indeed , it is thinkable that we can use the method to discover seminality of composers , or separate music eras and fads .
* a very interesting application of our program would be to select a plausible composer for a newly discovered piece of music of which the composer is not known .
in addition to such a piece , this experiment would require a number of pieces from known composers that are plausible candidates .
we would just run our program on the set of all those pieces , and see where the new piece is placed .
if it lies squarely within a cluster of pieces by composer such - and - such , then that would be a plausible candidate composer for the new piece . *
each run of our program is different even on the same set of data because of our use of randomness for choosing mutations in the quartet method
. it would be interesting to investigate more precisely how stable the outcomes are over different such runs .
* at various points in our program , somewhat arbitrary choices were made .
examples are the compression algorithms we use ( all practical compression algorithms will fall short of kolmogorov complexity , but some less so than others ) ; the way we transform the midi files ( choice of length of time interval , choice of note - representation ) ; the cost function in the quartet method .
other choices are possible and may or may not lead to better clustering .
ideally , one would like to have well - founded theoretical reasons to decide such choices in an optimal way .
lacking those , trial - and - error seems the only way to deal with them . *
the experimental results got decidedly worse when the number of pieces grew .
better compression methods may improve this situation , but the effect is probably due to unknown scaling problems with the quartet method or nonlinear scaling of possible similarities in a larger group of objects ( akin to the phenomenon described in the so - called `` birthday paradox '' : in a group of about two dozen people there is a high chance that at least two of the people have the same birthday ) .
inspection of the underlying distance matrices makes us suspect the latter . *
our program is not very good at dealing with very small data files ( 100 bytes or so ) , because significant compression only kicks in for larger files .
we might deal with this by comparing various sets of such pieces against each other , instead of individual ones .
we thank john tromp for useful discussions .
99 d. benedetto , e. caglioti , and v. loreto .
language trees and zipping , _ physical review letters _ , 88:4(2002 ) 048702 .
algorithm makes tongue tree , _ nature _ , 22 january , 2002 .
bennett , p. gcs , m. li , p.m.b .
vitnyi , and w. zurek .
information distance , _ ieee transactions on information theory _ , 44:4(1998 ) , 14071423 .
d. bryant , v. berry , p. kearney , m. li , t. jiang , t. wareham and h. zhang .
a practical algorithm for recovering the best supported edges of an evolutionary tree .
11th acm - siam symposium on discrete algorithms _
, 287296 , 2000 .
w. chai and b. vercoe .
folk music classification using hidden markov models .
_ proc . of international conference on artificial intelligence _ , 2001 .
r. dannenberg , b. thom , and d. watson .
a machine learning approach to musical style recognition , _ proc .
international computer music conference _ , pp .
344 - 347 , 1997 .
m. grimaldi , a. kokaram , and p. cunningham .
classifying music by genre using the wavelet packet transform and a round - robin ensemble .
technical report tcd - cs-2002 - 64 , trinity college dublin , 2002 .
t. jiang , p. kearney , and m. li . a polynomial time approximation scheme for inferring evolutionary trees from quartet topologies and its application .
_ siam j. computing _ , 30:6(2001 ) , 19421961 .
m. li , j.h .
badger , x. chen , s. kwong , p. kearney , and h. zhang . an information - based sequence distance and its application to whole mitochondrial genome phylogeny , _ bioinformatics _ , 17:2(2001 ) , 149154 .
m. li and p.m.b .
algorithmic complexity , pp .
376382 in : _ international encyclopedia of the social & behavioral sciences _ , n.j .
smelser and p.b .
baltes , eds . ,
pergamon , oxford , 2001/2002 .
m. li , x. chen , x. li , b. ma , p. vitnyi .
the similarity metric , _ proc .
14th acm - siam symposium on discrete algorithms _
, 2003 .
m. li and p.m.b .
_ an introduction to kolmogorov complexity and its applications _ , springer - verlag , new york , 2nd edition , 1997 .
p. scott .
music classification using neural networks , 2001 .
+ http://www.stanford.edu/class/ee373a/musicclassification.pdf shared information distance or software integrity detection , computer science , university of california , santa barbara , http://dna.cs.ucsb.edu/sid/ g. tzanetakis and p. cook , music genre classification of audio signals , _ ieee transactions on speech and audio processing _ , 10(5):293302 , 2002
.the 60 classical pieces used ( ` m ' indicates presence in the medium set , ` s ' in the small and medium sets ) [ cols="<,<,<",options="header " , ] | we present a fully automatic method for music classification , based only on compression of strings that represent the music pieces .
the method uses no background knowledge about music whatsoever : it is completely general and can , without change , be used in different areas like linguistic classification and genomics .
it is based on an ideal theory of the information content in individual objects ( kolmogorov complexity ) , information distance , and a universal similarity metric .
experiments show that the method distinguishes reasonably well between various musical genres and can even cluster pieces by composer . |
the authors thank s. raghu , s. lederer for helpful conversations . this work was supported by the doe , office of basic energy sciences , under contract no .
de - ac02 - 76sf00515
. 10 e. fradkin , s. a. kivelson , m. j. lawler , j. p. eisenstein , a. p. mackenzie , nematic fermi fluids in condensed matter physics . _ annu .
rev . condens .
matter phys . _
* 1 * , 153 - 178 ( 2010 )
. y. ando , k. segawa , s. komiya , a. n. lavrov , electrical resistivity anisotropy from self - organized one dimensionality in high - temperature superconductors .
lett . _ * 88 * , 137005 ( 2002 ) . c. howald , h. eisaki , n. kaneko , m. greven , a. kapitulnik , periodic density - of - states modulations in superconducting bi@xmath2sr@xmath2cacu@xmath2o@xmath56 .
b _ * 67 * , 014533 ( 2003 ) . v. hinkov
_ et al . _ , electronic liquid crystal state in the high - temperature superconductor yba@xmath2cu@xmath4o@xmath57 .
_ science _ * 319 * , 597 ( 2008 ) .
r. daou__et al . _
_ , broken rotational symmetry in the pseudogap phase of a high - t@xmath58 superconductor .
_ nature _ * 463 * , 519 ( 2010 ) .
s. a. kivelson _
_ , how to detect fluctuating stripes in the high - temperature superconductors .
phys . _ * 75 * , 1201 ( 2003 ) .
see i. r. fisher , l. degiorgi , z. x. shen , in - plane electronic anisotropy of underdoped ` 122 ' fe - arsenides revealed by measurements of detwinned single crystals .
phys . _ * 74 * , 124506 ( 2011 ) , and references there - in . see rafael m. fernandes , jrg schmalian , manifestations of nematic degrees of freedom in the magnetic , elastic , and superconducting properties of the iron pnictides .
_ supercond .
technol . _ * 25 * , 084005 ( 2012 ) , and references there - in .
chu , h .- h .
kuo , j. g. analytis , i. r. fisher , divergent nematic susceptibility in an iron arsenide superconductor .
_ science _ * 337 * , 710 - 712 ( 2012 ) .
h .- h kuo , maxwell c. shapiro , scott c. riggs , ian r. fisher , measurement of the elastoresistivity coefficients of the underdoped iron arsenide ba(fe@xmath59co@xmath60)@xmath2as@xmath2 .
b _ * 88 * , 085113 ( 2013 )
. h .- h kuo , i. r. fisher , effect of disorder on the resistivity anisotropy near the electronic nematic phase transition in pure and electron - doped bafe@xmath2as@xmath2 .
lett . _ * 112 * , 227001 ( 2014 ) .
m. yoshizawa _
et al . _ ,
structural quantum criticality and superconductivity in iron - based superconductor ba(fe@xmath61co@xmath1)@xmath2as@xmath2 .
_ j. phys .
jpn . _ * 81 * , 024604 ( 2012 ) .
a. e. bhmer _ et .
nematic susceptibility of hole - doped and electron - doped bafe@xmath2as@xmath2 iron - based superconductors from shear modulus measurements .
lett . _ * 112 * , 047001 ( 2014 ) .
et al . _ , direct observation of competition between superconductivity and charge density wave order in yba@xmath2cu@xmath4o@xmath62 . _
* 8 * , 871 ( 2012 ) .
g. ghiringhelli _
, long - range incommensurate charge fluctuations in ( y , nd)ba@xmath2cu@xmath4o@xmath63 .
_ science _ * 337 * , 821 ( 2012 ) .
e. blackburn _
x - ray diffraction observations of a charge - density - wave order in superconducting ortho - ii yba@xmath2cu@xmath4o@xmath64 single crystals in zero magnetic field .
lett . _ * 110 * , 137004 ( 2013 ) .
j. m. tranquada , b. j. sternlieb , j. d. axe , y. nakamura , s. uchida , evidence for stripe correlations of spins and holes in copper oxide superconductors .
_ nature _ * 375 * , 561 ( 1995 ) .
t. wu , h. mayaffre _ et al .
_ , magnetic - field - induced charge - stripe order in the high - temperature superconductor yba@xmath2cu@xmath4o@xmath65 .
_ nature _ * 477 * , 191 ( 2011 ) .
l. nie , g. tarjus , s. a. kivelson , quenched disorder and vestigial nematicity in the pseudogap regime of the cuprates .
_ * 111 * , 7980 ( 2014 ) .
s. a. kivelson , e. fradkin , v. j. emery , electronic liquid - crystal phases of a doped mott insulator .
_ nature _ * 393 * , 550 ( 1998 ) b. j. ramshaw _ et .
al . _ , a quantum critical point at the heart of high temperature superconductivity .
j. g. analytis _ et .
transport near a quantum critical point in bafe@xmath2(as@xmath61p@xmath1)@xmath2 .
phys . _ * 10 * , 194197 ( 2014 ) .
k. fujita _ et .
, simultaneous transitions in cuprate momentum - space topology and electronic symmetry breaking .
_ science _ * 344 * , 612 ( 2014 ) .
max a. metlitski , david f. mross , subir sachdev , t. senthil , are non - fermi - liquids stable to cooper pairing ? arxiv:1403.3694 ( unpublished ) s. lederer , y. schattner , e. berg , s. a. kivelson , enhancement of superconductivity near a nematic quantum critical point .
arxiv : 1406.1193 ( unpublished ) .
maier , d.j .
scalapino , pairing interaction near a nematic qcp of a 3-band cuo@xmath2 model .
arxiv:1405.5238 ( unpublished ) .
there are several different ways to define a nematic order parameter in terms of equilibrium electronic correlation functions .
correspondingly , there is an ambiguity in defining the nematic susceptibility .
thus the constants @xmath28 and @xmath29 not only depend on microscopic details , but they also depend on the precise microscopic definition we give to the nematic order parameter .
see supplementary material .
a. liam fitzpatrick , s. kachru , j. kaplan , s. raghu , non - fermi - liquid fixed point in a wilsonian theory of quantum critical metals .
b _ * 88 * , 125116 ( 2013 ) .
preliminary confirmation that the nematic susceptibility at criticality diverges as @xmath66 has been obtained from determinantal monte carlo calculations of a 2d metallic nematic quantum critical system .
s. lederer , y. shatner , e. berg , s.a .
kivelson , work in progress .
analytis , j .- h .
chu , r.d .
mcdonald , s. c. riggs , i.r .
fisher , enhanced fermi surface nesting in superconducting bafe@xmath2(as@xmath0p@xmath1)@xmath2 revealed by the de haas - van alphen effect .
lett . _ * 105 * , 207004 ( 2010 ) h. shishido _ et .
_ , evolution of the fermi surface of bafe@xmath2(as@xmath61p@xmath1)@xmath2 on entering the superconducting dome .
* 104 * , 057008 ( 2010 ) .
y. sun , s. thompson , t. nishida , _
strain effect in semiconductors : theory and device applications _
( springer , new york , 2010 ) .
chu , a. g. analytis , c. kucharczyk , i. r. fisher , determination of the phase diagram of the electron doped superconductor ba(fe@xmath0co@xmath1)@xmath2as@xmath2 .
b _ * 79 * , 014506 ( 2009 )
. part no .
pst 150/557 , from piezomechanik , munich , germany .
wk - xx-062tt-350 , general purpose strain gages - tee rosette , from vishay precision group .
14250 , general purpose adhesive epoxy , from devcon , u.s.a .
d. butkovicova _ et al .
_ , critical role of the sample preparation in experiments using piezoelectric actuators inducing uniaxial or biaxial strains . _ rev .
instr . _ * 84 * , 103902 ( 2013 ) .
jiun - haw chu _
et al . _ , in - plane resistivity anisotropy in an underdoped iron pnictide superconductor .
_ science _ * 329 * , 824 ( 2010 ) .
_ , possible origin of the nonmonotonic doping dependence of the in - plane resistivity anisotropy of ba(fe@xmath0t@xmath1)@xmath2as@xmath2 ( t = co , ni and cu ) .
b _ * 84 * , 054540 ( 2011 ) .
et al . _ , magneto - elastically coupled structural , magnetic and superconducting order parameters in bafe@xmath2(as@xmath0p@xmath1)@xmath2 . _
b _ * 86 * , 134507 ( 2012 ) .
e. c. blomberg _ et .
_ , sign - reversal of the in - plane resistivity anisotropy in hole - doped iron pnictides .
* 4 * , 1914 ( 2013 ) .
maria n. gastiasoro , i. paul , y. wang , p. j. hirschfeld , brian m. andersen , emergent defect states as a source of resistivity anisotropy in the nematic phase of iron pnictides .
_ * 113 * , 127001 ( 2014 ) .
wei bao _ et .
tunable ( @xmath67 , @xmath67)-type antiferromagnetic order in @xmath68-fe(te , se ) superconductors .
lett . _ * 102 * , 247001 ( 2009 ) .
t. m. mcqueen , _ et .
tetragonal - to - orthorhombic structural phase transition at 90 k in the superconductor fe@xmath69se _ phys .
* 103 * , 057002 ( 2009 ) .
d. s. fisher , random transverse field ising spin chains .
lett . _ * 69 * , 534 ( 1992 ) .
olexei motrunich , siun - chuon mau , david a. huse , daniel s. fisher , infinite - randomness quantum ising critical fixed points .
b _ * 61 * , 1160 ( 2000 ) .
e. w. carlson , k. a. dahmen , using disorder to detect locally ordered electron nematics via hysteresis .
* 2 * , 379 ( 2011 ) .
elastoresistivity relates changes of resistivity of a solid conductor to strains that it experiences .
since resistivity and strain are both 2nd rank tensors , the elastoresistivity is a 4th rank tensor , defined by the following expression : @xmath70 in the absence of a magnetic field , the onsager reciprocal relation holds for the electrical conductivity , and the associated symmetry properties of the resistivity and strain tensors allow us to more conveniently express them as six component vectors , such that @xmath71 where@xmath72 . using this voigt notation ,
the elastoresistivity is represented by a pseudo - second rank tensor . for a tetragonal crystal ,
the point group symmetry allows us to further simplify the elasotresistivity tensor : @xmath73 in the following sections , we describe how the elastoresistivity coefficients @xmath74 and @xmath35 are related to the nematic susceptibility for the @xmath19 and @xmath16 symmetry channels , and how these coefficients can be measured .
these ideas were first described in ref @xcite . in an experiment ,
the resistance of a sample is measured , not the resistivity .
the associated elastoresistance , i.e. the change of resistance of a sample due to deformation , is related to the elastoresistivity by : @xmath75 where @xmath76 is the separation of voltage leads and @xmath77 is the cross section area of the current path . in a simple metal , such as copper ,
the elastoresistivity coefficients are small , and the elastoresistance is dominated by the geometric factor ( the latter two terms in equation s5 ) . however , in semiconductors or strongly correlated materials the elastoresistivity coefficients can be much larger than the geometric effect @xcite . in the analysis described in the main text ,
the geometric factor is responsible for the small , temperature independent term @xmath78 .
single crystals were grown via standard self flux techniques , as described elsewhere @xcite .
thin bars were cut from larger crystals such that the tetragonal c - axis was perpendicular to the plane of the thin bars , and the a - axis aligned either along the length of the bars ( suitable for probing the b@xmath79 channel ) or at 45 degrees ( for b@xmath3 ) .
anisotropic strain was achieved by gluing these thin crystals to the side surface of a pzt piezoelectric stack@xcite , with the bars aligned along the principle @xmath52 and @xmath80 axes of the stack . by applying a positive voltage bias through
the voltage leads , the pzt stack expands along the long dimension and contracts along the transverse direction .
we define the direction along the long dimension as the @xmath80-direction and the transverse direction as @xmath52-direction ( see fig . 5 ) .
the resistance of pairs of mutually transverse samples was measured simultaneously while the voltage applied to the pzt stack was varied ( typically for three full hysteresis loops between -50 v and + 150 v ) . a photograph and illustration of the configuration of a typical experiment probing the differential elastoresistance of two such samples
is shown in fig .
5 . neglecting the geometric factors ( which
, as described above , contribute a small temperature - independent term to the elastoresistance ) , the change of resistance for a thin bar sample cut along the tetragonal @xmath81-axis and glued along the @xmath52-direction can be expressed by the following : @xmath82 where @xmath83 is the poisson s ratio of the pzt stack and @xmath84 is the effective poisson s ratio of the sample glued to the stack .
similarly , for the sample glued along the @xmath80-direction : @xmath85 by taking the difference of the change of the resistance of the two samples , one obtains : @xmath86 alternatively , by cutting the crystals in to bars oriented at 45 degrees with respect to the tetragonal @xmath81-axis , it can be easily shown via coordinate transformation that the difference of the change of the resistance of the two samples can be expressed by the following : @xmath87 where @xmath52 and @xmath80 refer to the principle axes of the pzt stack ( as illustrated in fig .
5 ) and not the crystallographic axes . since
@xmath88 and @xmath89 can be measured directly by the strain gauges glued on the back of the pzt stacks@xcite , we can directly extract the elastoresistivity coefficients by fitting the linear slope of @xmath90 as a function of strain .
an example of the linear fitting can be found in fig .
further details of the technique and the associated description of the elastoresistance can be found in ref .
@xcite and transverse ( @xmath91 elastoresistance for the @xmath16 symmetry channel.(b)@xmath92 as a function of anisotropic strain @xmath93 at several temperatures for the optimally doped ba(fe@xmath94ni@xmath95)@xmath2as@xmath2 .
black lines show linear fits for each temperature , from which the elatoresistivity coefficient @xmath35 is extracted .
strains are positive for extension , and negative for compression . ,
title="fig:",width=321][figs1 ] we use a standard two part epoxy@xcite to glue the samples on to the sidewall of the pzt stacks . it was found that the ability of the stack to transmit the strain to the sample depends on the choice of glue and the thickness and elastic properties of the sample@xcite . to confirm that the strain was fully transmitted to the samples , strain gauges were glued both on the back side of the stack and on top of a larger test crystal@xcite .
we found that the strain is essentially fully transmitted for sample thicknesses below approximately 100@xmath96 m ; measurements presented in this paper were made for samples with thicknesses between 50 and 100@xmath96 m .
the order parameter of a nematic phase is represented by a traceless symmetric tensor . in the case of the electronic nematic phase
, there is no single , unique representation , since any traceless symmetric tensor that measures the equilibrium electronic properties can represent the order parameter@xcite .
the dimensionless resistivity anisotropy is a traceless symmetric tensor , @xmath97 however , resistivity anisotropy does not purely measure the equilibrium electronic properties , but is determined by the anisotropy in the electronic structure _ and _ the anisotropy in the scattering rate ; the latter being a dynamical property of the system .
nevertheless , one can still express the true nematic order parameter @xmath98 as a function of the resistivity anisotropy @xmath99 and take a taylor expansion for small value of @xmath99 , for which the leading term is linear .
therefore , for small value of @xmath99 , which applies to our measurements ( which are all in the linear regime ) , it can be related to the true nematic order parameter via a proportionality constant @xmath99 = @xmath100 . in our experiment setup
, the strain can be considered as a symmetry breaking field conjugate to the electron nematic order parameter .
therefore we can define the nematic susceptibility as the quantity that relates the induced order parameter and symmetry breaking strain .
in particular , for a tetragonal crystal : @xmath101 where @xmath52 and @xmath80 refer here to the [ 100 ] and [ 010 ] crystal axes .
therefore the elastoresistivity coefficient is related to the bare nematic susceptibility via the same proportionality constant @xmath28 .
the proportionality constant @xmath28 depends on the microscopic details of the electronic structure , the precise definition given to the nematic order parameter , and , in some cases , the nature of the disorder scattering , and in general could be temperature and materials dependent .
however , two experimental observations of iron based superconductors suggest that @xmath28 is a constant within a wide range of temperatures and independent to a certain degree of disorder@xcite .
first , the temperature dependence of the elastoresistivity coefficient @xmath35 of the parent compounds bafe@xmath2as@xmath2 can be perfectly fitted by a curie - weiss function ( fig . 1 of main paper )
, which is exactly the same functional form of @xmath53 derived by a ginzburg - landau theory .
second , the elastoresistivity coefficient @xmath35 of bafe@xmath2as@xmath2 has been measured for crystals with different residual resistance ratios ( rrr ) , which is a measure of crystalline disorder @xcite .
it was found that @xmath35 is independent of the rrr of the sample .
a similar result is found for underdoped compositions of co- and ni - doped bafe@xmath2as@xmath2 which have the same neel temperature @xcites .
the perfect fitting to curie - weiss in these conditions strongly suggest that @xmath28 remains a constant over a wide range of temperatures and different degrees of disorder .
the main subject of our paper is the observation of a divergence of the differential elastoresistance that we have found for many different optimally doped compositions of fe - based superconductors .
however , beyond this main result several other features of the data are of interest in their own right and deserve additional comment . in this section
we particularly emphasize two aspects of the elastoresistance of optimally doped ba@xmath32k@xmath33fe@xmath2as@xmath2 . first ; whereas our measurements , described in the main text , reveal that the sign of the @xmath35 elastoresistivity coefficient is _ negative _ for co , ni and p substituted bafe@xmath2as@xmath2 , we find that it is _ positive _ for k - doped 122 .
second , the absolute magnitude of the coefficient is comparable for all four materials .
both of these observations warrant some additional commentary . for co , ni and p substituted 122 , the negative sign of m@xmath102 in the tetragonal phase ( for temperatures above the structural transition temperature , @xmath103 ) is consistent with the resistivity anisotropy that is observed for detwinned crystals in the orthorhombic state for temperatures _ below _
@xmath103 @xcite .
that is , the resistivity in the tetragonal phase _ decreases _ as the sample is stretched along the current direction . for 122 ,
the in - plane resistivity anisotropy has also been measured up to @xmath104 @xcite .
it was observed that for @xmath105 , the sign of resistivity anisotropy is the same as that of the parent compound and co - doped 122 .
but for @xmath106 , the sign of resistivity anisotropy was found to reverse ; i.e. the resistivity was found to be _ higher _ for the longer lattice constant direction .
our observation of a positive value for @xmath35 for optimal - doped 122 which has a potassium concentration @xmath107 , confirms this dramatic change in sign of the resistive anisotropy for sufficient hole doping .
significantly , our measurement is made in the tetragonal state , and therefore is not complicated by issues associated with the reconstruction of the fermi surface in the antiferromagnetic state , or by concerns associated with detwinning crystals in the orthorhombic state . despite the agreement in terms of the sign of the resistive anisotropy for sufficiently doped 122 ,
the relative magnitude of the anisotropy of detwinned underdoped compositions is quite different to the strain - induced elastoresistance near optimal doping .
the magnitude of the resistivity anisotropy of detwinned crystals of underdoped k - substituted 122 was found to be rather small @xcite .
this observation has been interpretted by some authors as evidence supporting the notion that the resistivity anisotropy derives from anisotropic scattering associated with lattice defects , which acquire an anisotropy due to the nematic fluctuations@xcite .
within such a scenario , defects in the feas planes are believed to have a stronger effect , and hence would naturally yield a larger resistivity ansiotropy , than defects that lie away from the feas plane ( as would be the case for k substitution ) , potentially accounting for the significantly lower resistivity anisotropy observed for detwinned underdoped 122 relative to 122 .
our measurements , however , reveal a comparable magnitude ( within a factor of two ) for @xmath35 of optimally doped co - substituted and k - substituted 122 , in contrast to the much larger difference in the resistivity anisotropy in the orthorhombic state for underdoped compositions .
clearly k substitution per se does not necessarily imply a small resistive anisotropy .
there are three possible explanations for the small anisotropy that was previously measured in the orthorhombic state for k - substituted bafe@xmath2as@xmath2 .
firstly , it is possible that the resistivity anisotropy has a small magnitude simply because it changes sign as a function of composition , and therefore must go through zero for some composition .
secondly , it is possible that the fs reconstruction in the antiferromagnetic state affects the resistivity anisotropy differently for electron and hole doped cases .
finally , it is also possible that the samples used for the study of underdoped compositions were not fully detwinned , although this seems less likely given the careful characterization of the twin populations by the authors @xcite . at present , it is not clear which of these possibilities is responsible for the small resistivity anisotropy of underdoped samples of 122 in the antiferromagnetic orthorhombic state . bearing this in mind , p - substitution gives another useful insight to the origin of the resistive anisotropy . beyond the residual resistivity ,
another good measure of the disorder affecting transport in the fe plane is the presence or absence of quantum oscillations .
p - substituted samples exhibit quantum oscillations across the phase diagram ( at least on the overdoped side@xcite , ) indicating a much lower elastic scattering rate than the other compounds we have discussed so far , for which quantum oscillations are not observed .
significantly , @xmath35 of the p - substituted samples is comparably large to that of co , ni and k doping , despite the lower elastic scattering rate , implying that the physical origin of the dominant contribution to the resistive anisotropy is unlikely to be associated with scattering from anisotropic defects , contrary to suggestions advanced in ref .
@xcite , even though such effects are necessarily present .
the observations described above serve to emphasize an important point ; that elastoresistance measurements for @xmath108 are the most relevant data to compare with theories of the resistivity anisotropy in the nematic phase , rather than comparison with transport anisotropies in the low t neel and/or orthorhombic state measured in samples detwinned by uni - axial stress .
there are several reasons , but primarily because ( a ) there are no concerns about degree of detwinning ( in the tetragonal phase the material is homogeneous , and homogeneously strained ) ; ( b ) there are no issues with back - folding of bands and fs reconstruction ; and ( c ) we have precise knowledge of the strain experienced by the material ( which is not known for detwinned samples , for which the amount of strain depends on the temperature dependent elastic modulus ) . in other words , in selecting transport data to compare with theoretical predictions for resistive anisotropy , elastoresistance measurements provide the more quantitatively controlled experimental perspective .
for reference , we show in figure 6 , the temperature - dependence of the resistivity for the same samples for which elastoresistance data are shown in the main text .
the data were taken for free - standing ( unstrained , and unattached ) crystals .
data are plotted as r / r(300k ) to eliminate uncertainty in geometric factors .
the `` 11 '' family of iron chacogenides ( ) provide an interesting example of the richness of the phase diagram of iron based superconductors . at the tellurium end of the phase diagram
, fe@xmath109te possesses a different antiferromagnetic ground state ; the antiferromagnetic order is a bicollinear order with an ordering wave - vector that is 45@xmath110 to that in iron pnictides@xcite . consequently the spontaneous strain developed in the symmetry breaking phase is @xmath93 ( b@xmath79 ) rather than @xmath111 ( b@xmath3 ) . on the other hand , at the selenium end of the phase diagram , fese only has a structural transition that develops a spontaneous b@xmath3 strain @xmath111 without any long range magnetic order@xcite .
the optimal doped fete@xmath32se@xmath33 we measured in this work has a @xmath112 and there is no sign of phase transitions other than the superconducting transition .
we have measured both @xmath113 ( b@xmath79 ) and @xmath35 ( b@xmath3 ) of fete@xmath32se@xmath33 , the results are presented in fig .
6 . from this figure
it can be clearly seen that the nematic suscptibility diverges in the b@xmath3 channel rather than the b@xmath79 channel . ) and @xmath31 of fete@xmath32se@xmath33 .
the data clearly show a divergence in the @xmath16 channel ( i.e. 2@xmath35 diverges ) , similar to the fe - arsenides , despite the fact that fete develop strain in the @xmath19 symmetry channel at the structural phase transition.,title="fig:",width=321][figs3 ]
as described in the main text , @xmath35 data were fit to a curie - weiss temperature dependence : @xmath34\ ] ] where @xmath114 gives the bare mean - field nematic critical temperature ( the temperature at which nematic order would occur if there were no coupling to the crystal lattice @xcite ) , and @xmath115 is the curie constant . for underdoped compositions , and for optimally doped bafe@xmath2(as@xmath36p@xmath37)@xmath2 , @xmath35 follows a perfect curie - weiss temperature dependence from room temperature down to @xmath14 and @xmath7 respectively . for optimally doped co , ni and k - substituted 122 , and for fete@xmath32se@xmath33
, the data can be described by curie - weiss behavior at higher temperatures , but not over the entire temperature range .
the range of temperatures over which this results in a reasonable description of the temperature - dependence of @xmath35 is shown by the linear fit ( red line ) to the inverse susceptibility @xmath116 in the lower panels of figures 2 and 3 in the main text .
associated fit parameters for compositions studied are listed in table 1 .
the quality of fit to curie - weiss behavior for optimally substituted bafe@xmath2(as@xmath36p@xmath37)@xmath2 can also be appreciated when the data are plotted on a log - log scale ( fig .
as can be seen , only the optimally p - substituted data ( red ) can be fitted linearly over the whole temperature range , with a slope @xmath117 ( solid black line ) . for all other optimally - doped samples , the data exhibit a downward deviation from curie - weiss behavior below a characteristic temperature that depends on the specific material .
elastoresistance of optimally doped bafe@xmath2(as@xmath36p@xmath37)@xmath2 .
figure shows @xmath118 $ ] v.s .
@xmath119 for all measured optimally - doped iron - based superconductors . @xmath120 and
@xmath12 were extracted from the curie - weiss fit@xcite . only the optimally p - substituted data ( red ) can be fitted linearly over the whole temperature range , with a slope @xmath117 ( solid black line ) .
note that for co , ni , and p substitution , @xmath121 is negative , whereas for k substitution and fete@xmath0se@xmath30 , it is positive .
, title="fig:",width=321][figs4 ] [ cols="^,^,^,^,^",options="header " , ]
here we analyze the properties of an electron nematic that derive primarily from general symmetry considerations . in particular , we discuss how the presence of quenched disorder affects the behavior of the nematic susceptibility upon approach to a nematic thermal or quantum critical point . admittedly , many aspects of the measurements we have reported can not be understood in this way as they depend on material - specific details and require an explicit understanding of the processes that contribute to the transport anisotropy , specifics of the particular type of quenched disorder involved , and of course an understanding of the underlying thermodynamic mechanism of nematicity .
nevertheless , this simple analysis serves to offer intuition for the effects of disorder in nematic systems .
it also reveals effects that have a remarkable similarity to the observed deviations from mean field curie - weiss behaviors of the elastoresistance for disordered optimally doped compounds presented in the main text .
we start by defining the local nematic order @xmath122 , which is even under inversion but odd under rotation by @xmath123 or reflection through the appropriate mirror plane .
the uniform thermal average of @xmath124 is the nematic order parameter , @xmath125 , where @xmath126 is proportional to the volume . in computing the susceptibility
, we will consider the effect of a symmetry breaking field , @xmath127 , which is conjugate to @xmath99 , and which thus represents the appropriate element of the strain tensor we define the coupling so that positive @xmath127 favors positive @xmath99 .
finally , the quenched disorder is represented by a random field @xmath128 which couples in the same way as @xmath127 to the nematic order parameter , but is assumed to have average strength zero and to be short - range correlated , @xmath129 and @xmath130 , so @xmath131 is the relevant measure of the strength of the disorder . * weak disorder
* : if the disorder is weak and the system is not too close to criticality , the effect of disorder on the nematic susceptibility can be computed perturbatively in the disorder strength , which yields a result of the form where @xmath134 , @xmath135 , and @xmath136 are all related to equilibrium properties of the disorder free system ( i.e. @xmath137 , for which formally exact expressions can be readily obtained in terms of various correlation functions of @xmath124 ) , and @xmath138 refers to higher order terms in powers of h and @xmath131 .
we have defined signs in these equations such that under typical circumstances @xmath139 , @xmath140 , and @xmath141 are all positive quantities . to obtain explicit expressions , we can consider the problem in the context of a landau - ginzberg free energy of the form @xmath142=\frac 1 2 \sum_{\vec r,\vec r^\prime } \phi(\vec
r ) g^{-1}(\vec r-\vec r^\prime)\phi_{\vec r^\prime } + \frac u 4 \sum_{\vec r } \phi^4_{\vec r } - \sum_{\vec r } [ h+b_{\vec r}]\phi(\vec r ) , \end{aligned}\ ] ] which we treat as a mean - field functional which we solve by minimizing @xmath143 with respect to @xmath124 . in the disordered phase ( where @xmath144 as @xmath145 ) , this gives rise to the expressions @xmath146^{(4-d)/2 } , \
\chi_3 = u[g_1]^4\end{aligned}\ ] ] where @xmath147^n\end{aligned}\ ] ] and where , for the usual reasons , @xmath148 .
this simple analysis already captures some qualitative features of the observed phenomena : a ) it shows that generically the effect of even weak disorder grows stronger upon approach to criticality , and conversely that at temperatures well away from criticality the effects can be negligible as far as the initial growth of critical correlations is concerned .
b ) moreover , it shows that generically the first effect of disorder is to slow the growth of the susceptibility relative to its expected behavior in the absence of disorder .
however , from this perturbative perspective , there is little reason to expect larger effects upon approach to a ( @xmath149 ) quantum critical point than upon approach to a classical ( @xmath150 ) critical point . * non - perturbative effects and quantum griffith phases *
: however , it is well known that disorder even weak disorder can produce non - perturbative corrections to the thermodynamic properties of a system , and even change the analytic structure of the free energy by producing so - called griffith singularities . for classical finite temperature systems , these griffith singularities tend to be unobservably weak . however , for quantum systems with a discrete order parameter these effects are highly amplified producing a new set of phenomena known as quantum griffith phases@xcite . in particular , based on symmetry
, we would tend to associate the universal critical phenomena associated with a continuous quantum phase transition to a nematic ground state with those of the transverse field ising model .
the nature of the quantum griffith phenomena associated with the transverse field ising model in the presence of random field disorder have been rigorously established in 1d@xcite , but scaling arguments which are successful in accounting for these results produce compelling results for higher dimensional problems as well@xcite . among other things ,
these arguments imply that for a range of transverse fields @xmath151 greater than a critical value , @xmath152 , the susceptibility diverges in the limit @xmath153 as @xmath154^{1 -\theta } \ { \rm with } \ \theta \sim ( h_\perp - h_{\perp , c})\end{aligned}\ ] ] even though the ground - state does not exhibit nematic order . at core , the enormous difference between classical and quantum griffith effects is illustrative of a more general tendency for the effects of disorder to be more dramatic on quantum critical phenomena than on their classical cousins . in classical statistical mechanics ,
a sum is performed over all position dependent field configurations with an appropriate boltzman weight , and quenched disorder implies the existence of point - like defects that locally favor one particular value of the field . in quantum statistical mechanics ,
the same sum is performed over field configurations that are a function of position and imaginary time , and quenched disorder implies the existence of line defects that are infinite in extent in the imaginary time direction .
thus , even rare impurities can have a large qualitative effect on the state of the system .
it seems to us that the fact that the observed effects of disorder on the nematic susceptibility appear strongest where the extrapolated nematic ordering temperature is close to t=0 reflects the enhanced sensitivity of quantum critical phenomena to disorder .
* non - equilibrium effects * : finally , it is important to note that non - equilibrium effects are unavoidable in any problem related to the random - field ising model .
the characteristic exponential slowing of dynamics upon approach to the critical temperature implies that necessarily at a temperature strictly larger than @xmath7 , on any accessible laboratory time scale the system will fail to achieve thermal equilibrium . at all lower temperatures
, glassy physics involving hysteresis , and noise must be expected@xcite .
the existence or not of `` aging '' effects reflecting this physics has not been explored in the present set of experiments . | by performing differential elastoresistivity measurements on a wider range of iron based superconductors , including electron doped ( ba(fe@xmath0co@xmath1)@xmath2as@xmath2 , ba(fe@xmath0ni@xmath1)@xmath2as@xmath2 ) , hole doped ( ba@xmath0k@xmath1fe@xmath2as@xmath2 ) , isovalent substituted pnictides ( bafe@xmath2(as@xmath0p@xmath1)@xmath2 ) and chalcogenides ( fete@xmath0se@xmath1 ) , we show that a divergent nematic susceptibility in the b@xmath3 symmetry channel appears to be a generic feature of optimally doped compositions .
122bafe@xmath2as@xmath2 122cafe@xmath2as@xmath2 122ba(fe@xmath0co@xmath1)@xmath2as@xmath2 122ba@xmath0k@xmath1fe@xmath2as@xmath2 327sr@xmath4ru@xmath2o@xmath5 * a growing body of evidence suggests the possibility of an intimate connection between electronic nematic phases @xcite and high - temperature superconductivity . at the moment it is unclear whether nematicity is an epi - phenomenon , or if there is a more fundamental causal relationship between nematic fluctuations and superconductivity . from an empirical perspective , strongly anisotropic
electronic phases are found in the underdoped regime of both cuprate @xcite and fe - based @xcite high - temperature superconductors .
for underdoped fe - based systems , recent measurements of the elastoresisistance @xcite and elastic moduli @xcite for the representative electron - doped system ba(fe@xmath0co@xmath1)@xmath2as@xmath2 reveal a divergence of the electronic nematic susceptibility upon approach to the tetragonal - to - orthorhombic structural phase transition , definitively establishing that the phase transition is driven by electronic correlation . for the cuprates ,
recent x - ray diffraction @xcite and nmr @xcite measurements have revealed evidence for short - range charge density wave order in `` underdoped '' crystals .
although details of the charge ordered state(s ) are still being established , these initial observations have at least motivated discussion of a possible `` vestigial '' nematic order @xcite .
perhaps significantly , in the phase diagrams of both families of compounds , optimal doping is located close to putative quantum critical points @xcite which potentially have a nematic character @xcite*. from a theoretical perspective , recent treatments indicate that nematic fluctuations due to proximity to a nematic quantum critical point can provide an effective pairing interaction , which , as a consequence of the @xmath6=0 nature of the fluctuations , enhance @xmath7 in all symmetry channels @xcite .
it is therefore of considerable interest to empirically establish whether nematic fluctuations are a characteristic feature of optimally doped high temperature superconductors . in the present paper ,
we show that this appears to be the case for fe - pnictide and chalcogenide superconductors by considering the representative materials ba(fe@xmath0co@xmath1)@xmath2as@xmath2 and ba(fe@xmath0ni@xmath1)@xmath2as@xmath2 ( i.e. electron - doped `` 122 '' ) , ba@xmath0k@xmath1fe@xmath2as@xmath2 ( hole - doped ) , bafe@xmath2(as@xmath0p@xmath1)@xmath2 ( isovalent substitution ) and fete@xmath0se@xmath1 ( `` 11 '' ) . furthermore , whilst the nematic susceptibility obeys a simple curie law for optimal bafe@xmath2(as@xmath0p@xmath1)@xmath2 , all other cases reveal sub - curie behavior , which we tentatively attribute to an enhanced sensitivity to disorder in a _
quantum _
critical regime .
nematic order couples linearly to anisotropic strain of the same symmetry .
consequently , the nematic susceptibility of a material can be measured by considering the electronic anisotropy that is induced by anisotropic in - plane strain . in the regime of infinitesimal strains ,
all forms of electronic anisotropy are linearly proportional .
hence , the rate of change of resistivity anisotropy with respect to anisotropic strain , defined in the limit of vanishing strain , is linearly proportional to the nematic susceptibility@xcite .
the proportionality constant depends on microscopic physics , but away from any quantum critical point does not contain any singular behavior .
consequently , the induced resistivity anisotropy reveals the essential divergence of the nematic susceptibility upon approach to a thermally - driven nematic phase transition , as we have demonstrated previously @xcite , and as is illustrated in figure 1 for the specific case of bafe@xmath2as@xmath2 .
elastoresistance of bafe@xmath2as@xmath2 . *
the data follow a curie - weiss behavior , which is the anticipated mean - field temperature dependence of the nematic susceptibility of a material approaching a thermally driven nematic phase transition @xcite .
upper panel shows @xmath8 , proportional to the nematic susceptibility @xmath9 .
black line shows the curie - weiss fit .
the remarkable quality of fit can be better appreciated by considering the inverse susceptibility , @xmath10 which is perfectly linear ( left axis of lower panel ; fit shown by red line ) , and the curie constant @xmath11 ( right axis of lower panel ) , which is independent of temperature .
the weiss temperature obtained from the curie - weiss fit , which gives the bare mean field nematic critical temperature , yields a value @xmath12 = 125 @xmath130.6 k. coupling to the lattice renormalizes the critical temperature , leading to a nematic - structural phase transition at @xmath14 = 134 k ( vertical gray line ) .
inset shows a photograph of mutually transverse crystals mounted on a pzt piezoelectric stack for differential elastoresistance measurements@xcite .
, width=321 ] for a tetragonal material , nematic order has either @xmath15 symmetry ( i.e. the @xmath16 irreducible representation of the @xmath17 point group , corresponding to nematic order oriented along a nearest - neighbor fe - fe bond the [ 110 ] or [ 110 ] crystal axes ) , or @xmath18 symmetry ( @xmath19 , corresponding to the [ 100 ] or [ 010 ] crystal axes ) . anisotropic strains with appropriate symmetry are then @xmath20 and @xmath21 respectively .
the strain - induced changes in resistivity can be described using the dimensionless elastoresistivity tensor , @xmath22 @xcite : @xmath23 where @xmath24 , @xmath25 etc . expressing the strain - induced differential elastoresistance _
i.e. _ @xmath26 \approx 2\big[\rho_{xx}-\rho_{yy}\big]/\big[[\rho_{xx}+\rho_{yy}\big]$ ] ) in terms of the elastoresistivity coefficients , and neglecting the geometric factors associated with straining the crystals@xcite , the corresponding components of the nematic susceptibility tensor are given by : @xmath27 where @xmath28 and @xmath29 are proportionality constants , which depend on microscopic physics@xcite . in figure 2
we show @xmath16 elastoresistance data for a range of optimally doped materials , including bafe@xmath2(as@xmath0p@xmath30)@xmath2 ( isovalently substituted ) , ba(fe@xmath0ni@xmath1)@xmath2as@xmath2 ( electron - doped ) , ba@xmath0k@xmath1fe@xmath2as@xmath2 ( hole - doped ) and fete@xmath0se@xmath1 . in all cases
, @xmath31 rises strongly with decreasing temperature , with comparably large values for each compound .
the observation of such an effect for this wide variety of ways of doping , including the case of the iron chalcogenide fete@xmath32se@xmath33 , is highly suggestive that a divergence of the nematic susceptibility in the @xmath16 channel is a generic feature of optimally doped fe - based superconductors .
this is our main result .
regardless of microscopic models , from a purely empirical perspective it is apparent that the optimally - doped superconductor is born out of an electronic state that is characterized by strongly fluctuating orientational ( nematic ) order in this specific symmetry channel .
this result is especially notable for fete@xmath32se@xmath33 given that the orientation of both the magnetic ordering wave vector and the in - plane component of the structural distortion of undoped fete is 45 degrees away from the orientation of those in the iron arsenide materials . to gain insight into the physical origin of the diverging elastoresistance , we fitted its temperature dependence by a curie - weiss functional form : @xmath34\ ] ] as can be seen from figure 2(a )
, the @xmath35 coefficient of optimally doped bafe@xmath2(as@xmath36p@xmath37)@xmath2 follows a perfect curie - weiss temperature dependence from room temperature down to @xmath7 ( below which temperature the resistance is zero ) , with a weiss temperature ( @xmath12 ) close to zero .
for all the other optimally doped systems , @xmath35 can be fit by curie - weiss behavior at higher temperatures , but shows a downward deviation below a characteristic temperature that is different for each of the materials .
the quality of fit to the curie - weiss functional form for bafe@xmath2(as@xmath36p@xmath37)@xmath2 in comparison to all of the other optimally doped compositions studied can be readily appreciated if the data are plotted on log axes .
as shown in figure s4@xcite , only the data for the p - substituted system can be fit by a single power law [ @xmath38 over the entire temperature range , yielding @xmath39 and @xmath40 k ) .
the small value of the weiss temperature observed for optimally doped bafe@xmath2(as@xmath36p@xmath37)@xmath2 motivates consideration of the nematic susceptibility in the context of a quantum critical point .
an ising nematic phase transition in an insulator would generally be expected to have dynamical exponent @xmath41 , so the effective dimensionality of the system would be @xmath42 = 3 + 1 . in the absence of quenched disorder , one would therefore anticipate mean field exponents ( up to possible logarithmic corrections ) at the quantum critical point .
as the materials in question are metallic , the situation is more complicated ( and probably even more interesting@xcite ) ; there are good reasons to expect that the hertz - millis paradigm breaks down in the case of electronic nematic order in metallic systems , for which landau damping associated with the gapless fermions leads to a larger value of the dynamical exponent , @xmath43 = 3 .
however , even taking this into account , the mean field behavior is still anticipated for the nematic susceptibility , consistent with the observed behavior of the @xmath16 elastoresistance of bafe@xmath2(as@xmath36p@xmath37)@xmath2@xcite .
in contrast to p - substitution , all of the other optimally doped compositions that we have studied reveal a downward deviation from curie - weiss behavior below some characteristic temperature . to gain insight into the physical origin of this effect ,
we consider the evolution of the @xmath16 elastoresistance as a function of composition for the specific case of co - doped bafe@xmath2as@xmath2 . panels ( a ) through ( d ) of figure 3 show the progression of @xmath31 as a function of co doping , from underdoped ( a , b ) , through optimal doping ( c ) , to the overdoped regime ( d ) . for underdoped compositions , @xmath31 can be well described by mean field curie - weiss t - dependence down to @xmath14 .
as has been previously established , this behavior is essentially independent of disorder , at least comparing pure bafe@xmath2as@xmath2 with different residual resistance ratios ( rrr ) and co and ni substituted crystals with the same @xmath44 @xcite .
figure 3(c ) shows @xmath31 for optimally doped ba(fe@xmath45co@xmath46)@xmath2as@xmath2 .
notice the significant deviation from mean - field behavior at low temperatures ; similar deviations are observed in optimally ni and k doped bafe@xmath47as@xmath2 and fete@xmath32se@xmath33 .
a similar effect has been observed in recent measurements of the sheer modulus via three - point bending experiments @xcite .
the deviations from curie - weiss behavior diminishes again as the doping is further increased , as can be seen in figure 3(d ) for the overdoped composition , for which the data can be fit to curie - weiss to a lower temperature than for optimal doping , and for which the magnitude of the deviations below @xmath48 45 k are smaller than for optimal doping . as can be see in figure 4 , the weiss temperature extracted from high - temperature fits to curie - weiss behavior crosses zero close to optimal doping .
the fact that deviations from curie - weiss behavior are largest for compositions for which @xmath12 approaches zero strongly suggests that the effect is associated with proximity to the putative nematic quantum phase transition .
significantly , p - substituted bafe@xmath2as@xmath2 is the `` cleanest '' of all of the known fe - pnictide and chalcogenide families , evidenced by the fact that quantum oscillations can be observed across the phase diagram @xcite .
the deviation from curie - weiss behavior for optimally doped compositions of all other dopants suggests that disorder plays an important role in the quantum critical regime , suppressing the nematic susceptibility below mean field expectations of a clean system .
all forms of quenched disorder produce locally anisotropic effective strains , which thus couple to the orientation of the nematic order ; in the language of statistical mechanics , this is `` random - field '' disorder@xcite .
in contrast , for commensurate magnetic order , disorder is of the random bond " variety , the effects of which are less profound .
certain generically expected effects of random field disorder are discussed in the supplemental material@xcite . specifically relevant
are the conclusions that : 1 ) for weak disorder , there is a perturbative regime away from criticality in which the effects of disorder are negligible , but they generically induce an increasingly strong reduction of the critical divergence of the susceptibility upon closer approach to the critical point .
2 ) while , perturbatively , classical and quantum critical phenomena have similar sensitivity to quenched disorder , close enough to criticality , where perturbation theory breaks down , the effects of disorder on quantum critical phenomena are qualitatively stronger , and likely give rise to quantum griffith phases .
3 ) close enough to criticality well - known non - equilibrium phenomena associated with the random field ising model should be apparent in the form of hysteresis , noise , and glassy aging effects . with the exception of point 3 ,
these expectations have some qualitative resemblance to the experimentally observed systematics . co@xmath1)@xmath2as@xmath2 . *
squares , circles and triangles indicate t@xmath49 , t@xmath50 , and t@xmath51 respectively .
stars indicate the bare mean - field nematic critical temperatures , @xmath12 , extracted from the curie - weiss fits of @xmath31 above the temperature at which disorder effects suppress the nematic susceptibility ( see main text ) .
as has been previously determined via longitudinal elastoresistance measurements , but established here by the full @xmath16 differential elastoresistance , the weiss temperature , @xmath12 goes through zero as a function of @xmath52 close to optimal doping @xcite .
color scale shows the magnitude of @xmath8 , which peaks close to optimal doping.,width=321 ] the temperature dependence of the nematic susceptibility can also be extracted from measurements of the elastic moduli @xcite .
the two measurements ( elastoresistance and elastic moduli ) reveal broad agreement , for example in terms of the curie - weiss t - dependence of @xmath53 for underdoped compositions of ba(fe@xmath0co@xmath1)@xmath2as@xmath2 , and also in terms of the deviation from curie - weiss behavior near optimal doping .
however , there is an important distinction in terms of the relative magnitude of the measured quantities as a function of doping . in particular , the normalized lattice softening @xmath54/c_{66}^0 $ ] extracted by boehmer _
et al . _
@xcite monotonically decreases in magnitude and extent in temperature as a function of @xmath52 .
however , as can be seen from inspection of figure 3 , the quantity @xmath55 initially increases with @xmath52 , peaking for optimally doped compositions .
the apparent enhancement of the elastoresistance 2@xmath35 over the softening of the elastic modulus for compositions near optimal doping is potentially related to renormalization of the quasiparticle effective mass in the quantum critical regime , as has been seen for instance in p - substituted bafe@xmath2as@xmath2 @xcite .
such a mass renormalization is an expected consequence of nematic quantum critical fluctuations@xcite .
significantly , it is precisely the low - energy quasiparticles that determine the elastoresistance that are also involved in the eventual superconductivity . |
single - walled carbon nanotube ( swcnt ) is a member from the carbon based nano - material family , which is characteristic for superior thermal conductivity.@xcite in many practical applications , the quasi - one - dimensional swcnt is ensembled in a bulk form , such as the swcnt mat ( or buckypaper).@xcite tubes in the swcnt mat are randomly orientated , leading to the lose of swcnt s advanced properties . for instance , the room temperature thermal conductivity in the swcnt mat is typically in the range of @xmath0 to 10 wm@xmath1k@xmath1 , which is several orders lower than the swcnt.@xcite besides the randomly orientated swcnt mat , there is a particular mat structure , with all tubes aligning parallel with each other .
this structure is termed as swcnt forest , or swcnt crystalline .
the vertically aligned swcnt forest has promising practical applications such as supercapacitor electrodes for compact energy - storage devices , since it retains most mechanical , thermal , and electronic properties of the swcnt.@xcite during the development of the swcnt forest , the density used to be a bottleneck for its practical applications , because most advanced properties of the swcnt forest will disappear if its density is too low.@xcite great progress has been achieved in recent years toward the production of dense swcnt forest .
futaba et al .
fabricated a high - density swcnt forest by using the zipping effect of liquids to draw tubes together.@xcite the density of the swcnt forest is increased to @xmath2 @xmath3 by zhong et al .
using the point - arc microwave plasma chemical vapor deposition.@xcite in 2010 , a general catalytic chemical vapor deposition ( cvd ) design is developed to synthesize aligned forests of carbon nanotube with an area density above @xmath4 @xmath3 , which can be further increased by using narrower tubes.@xcite this value is already very close to the theoretical maximum density in a compact array . during the application of dense swcnt forests in the electronic device such as the supercapacitor electrodes , the thermal management becomes a key factor for its breakdown due to high power density . in this point of view , it is important to investigate the thermal transport capability of the swcnt forest.@xcite the swcnt forest inherits the good thermal transport properties in its axial direction from the swcnt . its thermal conductivity is around 1750 - 5800 wm@xmath1k@xmath1 in the axial direction as estimated in the experiment.@xcite the 3-omega measurements obtained the axial thermal conductivity of the swcnt forest to be 83 wm@xmath1k@xmath1 at 323 k , which should be higher with sample quality improvement.@xcite zhang et al
. demonstrated that the room temperature axial thermal conductivity in swcnt forest is about 450 wm@xmath1k@xmath1 , and this value can be increased linearly by increasing forest density.@xcite besides the axial thermal transport , the inter - tube thermal conductivity is also very helpful to pump thermal energy out off the electronic device .
although the swcnt forest is able to deliver heat along its axial direction very fast , the inter - tube thermal conductivity in the swcnt forest is extremely low due to the weak van der waals inter - tube interaction . [
1.0 ] and the relaxation time of the heat bath is @xmath5 ps . the top left inset shows the linear fitting for the profile in @xmath6 $ ] , giving a temperature gradient @xmath7 .
the right inset shows the linear fitting for the profile in @xmath8 $ ] , giving a temperature gradient @xmath9 .
these two temperature gradients ( @xmath7 and @xmath9 ) are averaged to give the temperature gradient in the computation of the thermal conductivity using the fourier law .
bottom configuration shows the temperature distribution in the swcnt forest .
the color is with respective to the temperature of each carbon atom.,title="fig:",width=302 ] to improve the situation , significant effort has been made to increase the inter - tube thermal conductivity.@xcite it was found that the thermal conductivity of the nanotube forest can be enhanced considerably by increasing its density.@xcite basically , the low inter - tube thermal conductivity is mainly because the van der waals interaction between two swcnts is too weak . to enhance the inter - tube interaction , the ion irradiation technique was applied to introduce displacements between neighboring tubes , so the inter - tube thermal conductivity can be increased.@xcite another straightforward solution to enhance the inter - tube van der waals interaction is to compress the swcnt forest using mechanical engineering approach .
this forms the main scope of present work . in this paper , we perform molecular dynamics simulations to examine the compression effect on the inter - tube thermal conductivity of the swcnt forest .
we find that the thermal conductivity increases linearly with increasing compression before the swcnt forest is buckled , because the inter - tube interaction is enhanced by compression .
however , the thermal conductivity decreases quickly with further compression after the buckling of the swcnt forest , where the cross section of the swcnt changes from circular into ellipse .
it is because the twisting ( tw ) mode is depressed in the buckled swcnt forest , while the radial breath mode ( rbm ) becomes a favorable vibration mode .
the rbm has a poor heat delivery capability , leading to the decrease of the thermal conductivity with increasing compression in the buckled swcnt forest .
insets ( left to right ) show local configurations for swcnt forest at compression with strain @xmath10 -0.05 , -0.07 , and -0.08 , respectively .
colors are with respective to the potential of each atom.,title="fig:",width=302 ] [ 1 ]
we study the the inter - tube thermal conductivity , which captures the ability of transporting heat energy from a swcnt to its neighboring tubes in the forest .
our study focuses on the swcnt forest that is constructed with armchair swcnt ( 8 , 8) .
all swcnts are closely packed and parallel to each other in the axial direction ( see the bottom configuration in fig . [ fig_dtdx ] ) .
there are 26 swcnts aligned in the longitudinal direction and 4 swcnts in the lateral direction .
the length of an individual swcnt is 14.8 , which forms the thickness of the swcnt forest .
the thermal transport is simulated by the direct molecular dynamics simulation method implemented in the lammps package , while the ovito package was used for visualization in this section.@xcite the c - c interactions in the swcnt forest are described by the adaptive intermolecular reactive empirical bond order ( airebo ) potential.@xcite the distance cut - off for the lennard - jones potential in the airebo is chosen as 1.02 nm .
the standard newton equations of motion are integrated in time using the velocity verlet algorithm with a time step of 1 fs .
periodic boundary conditions are applied in all directions .
the thermal current across the system is driven by frequently moving kinetic energy from the cold region to the hot region.@xcite this energy transfer is accomplished via scaling the velocity of atoms in the hot or cold temperature - controlled regions .
the total thermal current pumped into the hot region ( or pumped out from the cold region ) is @xmath11 .
we have introduced a current parameter @xmath12 to measure the energy amount to be aggregated per unit time .
we note that the total current @xmath13 is the _ eflux _ parameter in the _ fix heat _ command in lammps .
the physical meaning of the parameter @xmath12 is the ratio between the aggregated energy and the kinetic energy of each atom , i.e @xmath12 measures the thermal current for each atom with respect to its kinetic energy.@xcite we have chosen a small parameter @xmath14 .
a small @xmath12 is very important for a successful simulation of inter - tube thermal transport , because the inter - tube thermal conductivity is so weak that atoms in cold regions will be frozen if their kinetic energies are pumped out too fast .
the velocity scaling operation is performed every @xmath15 ps , which can be regarded as the relaxation time of this particular heat bath .
we choose @xmath5 ps in all simulations .
after thermalization for a sufficiently long time , a temperature profile is established .
[ fig_dtdx ] shows the temperature profile for the swcnt forest at room temperature .
the two insets show that the left middle region with relative axial position in @xmath16 $ ] and the right middle region within @xmath17 $ ] are linearly fitted to extract two temperature gradients . these two temperature gradients are averaged to give the final temperature gradient @xmath18 .
the thermal conductivity is obtained through the fourier law , @xmath19 where a factor of @xmath20 comes from the fact that heat current flows in two opposite directions .
the difference between the left and right temperature gradients is used to estimate the numerical error in the thermal conductivity .
fig . [ fig_stress_strain ] shows the stress - strain relation for the swcnt forest during mechanical compression at room temperature .
the structure is compressed by uniformly deforming the structure in longitudinal direction with a strain rate of @xmath21 s@xmath1 , which is a typical value in md simulations as shown in previous works .
@xcite the structure is allowed to fully relax in transverse directions .
insets illustrate representative local configurations of the swcnt forest , which is compressed at strains of @xmath10 -0.05 , -0.07 , and -0.08 , respectively .
the buckling of swcnts in the forest starts at @xmath22 . after buckling
, the cross section of the swcnt turns into ellipse and the stress concentration becomes obvious at the two narrow corners of the ellipse , which is indicated by more bright color .
phonon modes are the solely thermal energy carrier in the lattice thermal transport simulated in present work , so it is necessary to illustrate some associated phonon modes .
[ fig_u ] shows the eigen vector of five typical phonon modes in the swcnt ( 8 , 8) .
the blue arrow on top of each atom represents the vibrational displacement of each carbon atom in the corresponding phonon mode .
the first four panels represent the four acoustic modes , i.e , the y - direction transverse acoustic ( tay ) , the z - direction transverse acoustic ( taz ) , the longitudinal acoustic ( la ) , and the tw mode .
the last panel shows the rbm . to examine the compression effect on the thermal conductivity of the swcnt forest , we present the compression dependence of the thermal conductivity in fig . [ fig_thermal_conductivity ] .
the swcnt forest is compressed to a given strain value prior to the heat transport simulation .
the thermal conductivity is then simulated with the non - equilibrium molecular dynamics method as outlined in previous section .
we pay special attention to the modulation of the thermal conductivity by the buckling phenomenon at @xmath22 .
it shows that the mechanical compression can enhance the thermal conductivity linearly before the occurrence of buckling phenomenon at @xmath22 , which is different from the robust thermal conductivity in the graphene under compression.@xcite left top inset illustrates an instant local configuration of the swcnt forest at compression with @xmath23 before buckling .
the red arrow on top of each atom represents the velocity of the atom . after comparing this configuration with the eigen vector of the tw mode in fig .
[ fig_u ] , one finds that the vibrations of most swcnts are similar as the eigen vector of the tw phonon mode ; i.e , the tw mode plays an important role for heat transport in the forest .
it also implies that the tw - like movement has lower energy threshold , as the interface between two circular swcnts is very smooth . as a result , the tw - like movement is easy to be excited . [ 1.0 ] [ 1.0 ] .
red arrows attached on each atom denote the velocity of carbon atoms .
the texts tw and ta imply that swcnts are involving in movements similar as the tw mode or the ta mode .
right bottom inset shows an instant local configuration for the swcnt forest at compression @xmath24 , where the cross section of the swcnt becomes ellipse .
the movements of swcnts are similar as the tw mode or the rbm .
lines are guide to the eyes.,title="fig:",width=302 ] on contrary , fig .
[ fig_thermal_conductivity ] also shows that , for larger compression with @xmath25 , the thermal conductivity decreases quickly with increasing compression .
right bottom inset illustrates an instant local configuration of the swcnt forest at compression with @xmath24 after buckling phenomenon . as compared with the local configuration before buckling , a distinct feature in the buckled structure
is that most swcnts are involved in a motion that is similar as the eigen vector of the rbm shown in fig .
[ fig_u ] .
the tw - like movement has been strongly depressed , because swcnts with ellipse cross section are difficult to twist , as a result of the effectively rough interface between two neighboring ellipse swcnts .
we have performed classical molecular dynamics simulations to study the thermal conductivity of the compressed swcnt forest .
the mechanical compression can enhance the thermal conductivity before the buckling of the swcnt in the forest , but it weakens the thermal conductivity after the buckling happens . the origin for this thermal conductivity manipulation is discussed based on the interplay between the tw mode ( good heat carrier ) and the rbm ( poor heat carrier ) .
the work is supported by the recruitment program of global youth experts of china and the start - up funding from shanghai university .
the author thanks ai - jun li in shu for valuable comments .
41 [ 1]#1 [ 1]`#1 ` [ 2]#2 [ 1]#1 [ 1]http://arxiv.org / abs/#1 [ ] [ 1]http://dx.doi.org/#1 [ ] [ 1]pmid:#1 [ ] [ 2]#2 , , , , . .
, , , , , , et al . .
, , , , , . .
, , , , . .
, , , , , , et al . . ; : . , .
, , , , , , et al . .
, , , , , , et al . .
; : . , , .
, , , , , . .
, , , , . .
; : . , , , , , , et al . .
, , , , , , et al . .
, , , , , , et al . .
, , , , , , et al . . ; : .
, , , , . .
, , , , , . .
; : . , , .
, , , , , , et al . .
; ( ): . , .
( ): . , , .
, , , , . .
, , , , . .
, , , , , , et al . .
, , . . ; ( ): . ,
( ): . , , .
, , , , . .
, , , . . ; ( ): | we perform classical molecular dynamics simulations to investigate the mechanical compression effect on the thermal conductivity of the single - walled carbon nanotube ( swcnt ) forest , in which swcnts are closely aligned and parallel with each other .
we find that the thermal conductivity can be linearly enhanced by increasing compression before the buckling of swcnt forests , but the thermal conductivity decreases quickly with further increasing compression after the forest is buckled .
our phonon mode analysis reveals that , before buckling , the smoothness of the inter - tube interface is maintained during compression , and the inter - tube van der waals interaction is strengthened by the compression .
consequently , the twisting - like mode ( good heat carrier ) is well preserved and its group velocity is increased by increasing compression , resulting in the enhancement of the thermal conductivity .
the buckling phenomenon changes the circular cross section of the swcnt into ellipse , which causes effective roughness at the inter - tube interface for the twisting motion . |
nowadays , many engineering applications can be posed as convex quadratic problems ( qp ) .
several important applications that can be modeled in this framework such us model predictive control for a dynamical linear system @xcite and its dual called moving horizon estimation @xcite , dc optimal power flow problem for a power system @xcite , linear inverse problems arising in many branches of science @xcite or network utility maximization problems @xcite have attracted great attention lately . since the computational power has increased by many orders in the last decade ,
highly efficient and reliable numerical optimization algorithms have been developed for solving the optimization problems arising from these applications in very short time .
for example , these hardware and numerical recent advances made it possible to solve linear predictive control problems of nontrivial sizes within the range of microseconds and even on hardware platforms with limited computational power and memory @xcite .
the theoretical foundation of quadratic programming dates back to the work by frank & wolfe @xcite . after the publication of the paper @xcite many numerical algorithms
have been developed in the literature that exploit efficiently the structure arising in this class of problems .
basically , we can identify three popular classes of algorithms to solve quadratic programs : active set methods , interior point methods and ( dual ) first order methods .
_ active set methods _ are based on the observation that quadratic problems with equality constraints are equivalent to solving a linear system .
thus , the iterations in these methods are based on solving a linear system and updating the active set ( the term active set refers to the subset of constraints that are satisfied as equalities by the current estimate of the solution ) .
active set general purpose solvers are adequate for small - to - medium scale quadratic problems , since the numerical complexity per iteration is cubic in the dimension of the problem .
matlab s _ quadprog _ function implements a primal active set method .
dual active set methods are available in the codes @xcite .
_ interior point methods _ remove the inequality constraints from the problem formulation using a barrier term in the objective function for penalizing the constraint violations .
usually a logarithmic barrier terms is used and the resulting equality constrained nonlinear convex problem is solved by the newton method . since the iteration complexity grows also cubically with the dimension , interior - point solvers are also the standard for small - to - medium scale qps .
however , structure exploiting interior point solvers have been also developed for particular large - scale applications : e.g. several solvers exploit the sparse structure of the quadratic problem arising in predictive control ( cvxgen @xcite , forces @xcite ) .
a parallel interior point code that exploits special structures in the hessian of large - scale structured quadratic programs have been developed in @xcite . _
first order methods _ use only gradient information at each iterate by computing a step towards the solution of the unconstrained problem and then projecting this step onto the feasible set .
augmented lagrangian algorithms for solving general nonconvex problems are presented in the software package lancelot @xcite .
for convex qps with simple constraints we can use primal first order methods for solving the quadratic program as in @xcite . in this case
the main computational effort per iteration consists of a matrix - vector product . when the projection on the primal feasible set is hard to compute , an alternative to primal first order methods is to use the lagrangian relaxation to handle the complicated constraints and then to apply dual first order algorithms for solving the dual .
the computational complexity certification of first order methods for solving the ( augmented ) lagrangian dual of general convex problems is studied e.g. in @xcite and of quadratic problems is studied in @xcite . in these methods the main computational effort consists of solving at each iteration a lagrangian qp problem with simple constraints for a given multiplier , which allows us to determine the value of the dual gradient for that multiplier , and then update the dual variables using matrix - vector products .
for example , the toolbox fiordos @xcite auto - generates code for primal or dual fast gradient methods as proposed in @xcite .
the algorithm in @xcite dualizes only the inequality constraints of the qp and assumes available a solver for linear systems that is able to solve the lagrangian inner problem .
however , both implementations @xcite do not consider the important aspect that the lagrangian inner problem can not be solved exactly in practice .
the effect of inexact computations in dual gradient values on the convergence of dual first order methods has been analyzed in detail in @xcite .
moreover , most of these papers generate approximate primal solutions through averaging @xcite . on the other hand , in practice
usually the last primal iterate is employed , since in practice these methods converge faster in the primal last iterate than in a primal average sequence .
these issues motivate our work here . _
contributions_. in this paper we analyze the computational complexity of several ( augmented ) dual first order methods implemented in duquad for solving convex quadratic problems . contrary to most of the results from the literature @xcite , our approach allows us to use inexact dual gradient information ( i.e. it allows to solve the ( augmented ) lagrangian inner problem approximately ) and therefore is able to tackle more general quadratic convex problems and to solve practical applications .
another important feature of our approach is that we provide also complexity results for the primal latest iterate , while in much of the previous literature convergence rates in an average of primal iterates are given @xcite .
we derive in a unified framework the computational complexity of the dual and augmented dual ( fast ) gradient methods in terms of primal suboptimality and feasibility violation using inexact dual gradients and two types of approximate primal solutions : the last primal iterate and an average of primal iterates . from our knowledge
this paper is the first where both approaches , dual and augmented dual first order methods , are analyzed uniformly .
these algorithms are also implemented in the efficient programming language c in duquad , and optimized for low iteration complexity and low memory footprint .
the toolbox has a dynamic matlab interface which make the process of testing , comparing , and analyzing the algorithms simple .
the algorithms are implemented using only basic arithmetic and logical operations and thus are suitable to run on low cost hardware .
the main computational bottleneck in the methods implemented in duquad is the matrix - vector product .
therefore , this toolbox can be used for solving either qps on hardware with limited resources or sparse qps with large dimension .
_ contents_. the paper is organized as follows . in section [ sec_pf ]
we describe the optimization problem that we solve in duquad . in section [ sec_duquad ]
we describe the the main theoretical aspects that duquad is based on , while in section [ numerical_tests ] we present some numerical results obtained with duquad . _
notation_. for @xmath0 denote the scalar product by @xmath1 and the euclidean norm by @xmath2 .
further , @xmath3_x$ ] denotes the projection of @xmath4 onto convex set @xmath5 and @xmath6_x\|$ ] its distance . for a matrix
@xmath7 we use the notation @xmath8 for the spectral norm .
in duquad we consider a general convex quadratic problem ( qp ) in the form : @xmath9 where @xmath10 is a convex quadratic function with the hessian @xmath11 , @xmath12 , @xmath13 is a simple compact convex set , i.e. a box @xmath14 $ ] , and @xmath15 is either the cone @xmath16 or the cone @xmath17 . note that our formulation allows to incorporate in the qp either linear inequality constraints @xmath18 ( arising e.g. in sparse formulation of predictive control and network utility maximization ) or linear equality constraints @xmath19 ( arising e.g. in condensed formulation of predictive control and dc optimal power flow ) .
in fact the user can define linear constraints of the form : @xmath20 and depending on the values for @xmath21 and @xmath22 we have linear inequalities or equalities . throughout the paper
we assume that there exists a finite optimal lagrange multiplier @xmath23 for the qp and it is difficult to project on the feasible set of problem : @xmath24 therefore , solving the primal problem approximately with primal first order methods is numerically difficult and thus we usually use ( augmented ) dual first order methods for finding an approximate solution for . by moving the complicating constraints @xmath25 into the cost via lagrange multipliers
we define the ( augmented ) dual function : @xmath26 where @xmath27 denotes the ( augmented ) lagrangian w.r.t .
the complicating constraints @xmath28 , i.e. : @xmath29 where the regularization parameter @xmath30 .
we denote @xmath31 and observe that : @xmath32_{{\mathcal{k } } } & \text{if } \;\ ; \rho>0 \\ 0 & \text{if } \;\ ; \rho=0 .
\end{cases}\ ] ] using this observation in the formulation , we obtain : @xmath33 in order to tackle general convex quadratic programs , in duquad we consider the following two options : * case 1 * : if @xmath34 , i.e. @xmath35 has the smallest eigenvalue @xmath36 , then we consider @xmath37 and recover the ordinary lagrangian function . *
case 2 * : if @xmath11 , i.e. @xmath35 has the smallest eigenvalue @xmath38 , then we consider @xmath39 and recover the augmented lagrangian function . our formulation of the ( augmented ) lagrangian and the previous two cases allow us to thereat in a unified framework both approaches , dual and augmented dual first order methods , for general convex qps .
we denote by @xmath40 the optimal solution of the _ inner problem _ with simple constraints @xmath41 : @xmath42 note that for both cases described above the ( augmented ) dual function is differentiable everywhere .
moreover , the gradient of the ( augmented ) dual function @xmath43 is @xmath44-lipschitz continuous and given by @xcite : @xmath45 for all @xmath46 .
since the dual function has lipschitz continuous gradient , we can derive bounds on @xmath47 in terms of a linear and a quadratic model ( the so - called _ descent lemma _ ) @xcite : @xmath48 - d_\rho(\lambda ) \leq \frac{l_{\text{d}}}{2 } \|\mu - \lambda\|^2 \quad \forall \lambda , \mu \in { \mathbb{r}^}p.\end{aligned}\ ] ] descent lemma is essential in proving convergence rate for first order methods @xcite . since we assume the existence of a finite optimal lagrange multiplier for , strong duality holds and thus the _ outer problem
_ is smooth and satisfies : @xmath49 where @xmath50 note that , in general , the smooth ( augmented ) dual problem is not a qp , but has simple constraints .
we denote a primal optimal solution by @xmath51 and a dual optimal solution by @xmath23 .
we introduce @xmath52 as the set of optimal solutions of the smooth dual problem and define for some @xmath53 the following finite quantity : @xmath54 in the next section we present a general first order algorithm for convex optimization with simple constraints that is used frequently in our toolbox . in this section
we present a framework for first order methods generating an approximate solution for a smooth convex problem with simple constraints in the form : @xmath55 where @xmath56 is a convex function and @xmath5 is a simple convex set ( i.e. the projection on this set is easy ) .
additionally , we assume that @xmath57 has lipschitz continuous gradient with constant @xmath58 and is strongly convex with constant @xmath59 .
this general framework covers important particular algorithms @xcite : e.g. gradient algorithm , fast gradient algorithm for smooth problems , or fast gradient algorithm for problems with smooth and strongly convex objective function .
thus , we will analyze the iteration complexity of the following general first order method that updates two sequences @xmath60 as follows : where @xmath61 is the parameter of the method and we choose it in an appropriate way depending on the properties of function @xmath57 .
more precisely , @xmath61 can be updated as follows : * gm * : in the gradient method @xmath62 , where @xmath63 for all @xmath64 .
this is equivalent with @xmath65 for all @xmath64 . in this case @xmath66 and thus we have the classical gradient update : @xmath67_x$ ] . *
fgm * : in the fast gradient method for smooth convex problems @xmath68 , where @xmath69 and @xmath70 . in this case we get a particular version of nesterov s accelerated scheme @xcite that updates two sequences @xmath71 and has been analyzed in detail in @xcite .
* * fgm**@xmath72 : in fast gradient algorithm for smooth convex problems with strongly convex objective function , with constant @xmath73 , we choose @xmath74 for all @xmath64 . in this case
we get a particular version of nesterov s accelerated scheme @xcite that also updates two sequences @xmath71 .
the convergence rate of algorithm * fom*(@xmath75 ) in terms of function values is given in the next lemma : @xcite [ lemma_sublin_dfg ] for smooth convex problem assume that the objective function @xmath57 is strongly convex with constant @xmath76 and has lipschitz continuous gradient with constant @xmath58 .
then , the sequences @xmath77 generated by algorithm * fom*(@xmath75 ) satisfy : @xmath78 where @xmath79 , with @xmath80 the optimal set of , and @xmath81 is defined as follows : @xmath82 thus , algorithm * fom * has linear convergence provided that @xmath83 .
otherwise , it has sublinear convergence .
in this section we describe an inexact dual ( augmented ) first order framework implemented in duquad , a solver able to find an approximate solution for the quadratic program . for a given accuracy @xmath84
, @xmath85 is called an @xmath86-_primal solution _ for problem if the following inequalities hold : @xmath87 the main function in duquad is the one implementing the general algorithm * fom*. note that if the feasible set @xmath88 of is simple , then we can call directly * fom*(@xmath89 ) in order to obtain an approximate solution for . however , in general the projection on @xmath90 is as difficult as solving the original problem . in this case
we resort to the ( augmented ) dual formulation for finding an @xmath86-primal solution for the original qp .
the main idea in duquad is based on the following observation : from we observe that for computing the gradient value of the dual function in some multiplier @xmath91 , we need to solve exactly the inner problem ; despite the fact that , in some cases , the ( augmented ) lagrangian @xmath92 is quadratic and the feasible set @xmath93 is simple in , this inner problem generally can not be solved exactly .
therefore , the main iteration in duquad consists of two steps : * step 1 * : for a given inner accuracy @xmath94 and a multiplier @xmath95 solve approximately the inner problem with accuracy @xmath94 to obtain an approximate solution @xmath96 instead of the exact solution @xmath97 , i.e. : @xmath98 in duquad , we obtain an approximate solution @xmath96 using the algorithm * fom*(@xmath99 ) .
from , lemma [ lemma_sublin_dfg ] we can estimate tightly the number of iterations that we need to perform in order to get an @xmath94-solution @xmath96 for : the lipschitz constant is @xmath100 , the strong convexity constant is @xmath101 ( provided that e.g. @xmath19 ) and @xmath102 ( the diameter of the box set @xmath93 ) .
then , the number of iterations that we need to perform for computing @xmath96 satisfying can be obtained from . *
step 2 * : once an @xmath94-solution @xmath96 for was found , we update at the outer stage the lagrange multipliers using again algorithm * fom*(@xmath103 ) . note that for updating the lagrange multipliers we use instead of the true value of the dual gradient @xmath104 , an approximate value given by : @xmath105 . in @xcite it has been proved separately , for dual and augmented dual first order methods , that using an appropriate value for @xmath94 ( depending on the desired accuracy @xmath86 that we want to solve the qp ) we can still preserve the convergence rates of algorithm * fom*(@xmath103 ) given in lemma [ lemma_sublin_dfg ] , although we use inexact dual gradients . in the sequel ,
we derive in a unified framework the computational complexity of the dual and augmented dual ( fast ) gradient methods . from our knowledge , this is the first time when both approaches , dual and augmented dual first order methods , are analyzed uniformly .
first , we show that by introducing inexact values for the dual function and for its gradient given by the following expressions : @xmath106 then we have a similar descent relation as in given in the following lemma : given @xmath107 such that holds , then based on the definitions we get the following inequalities : @xmath108 - d_\rho(\lambda ) \leq\ !
l_{\text{d } } \|\mu - \lambda\|^2 + \ !
2 { \epsilon_{\text{in}}}\quad \forall \lambda,\mu \!\in\ ! { \mathbb{r}^}p.\end{aligned}\ ] ] from the definition of @xmath109 , and it can be derived : @xmath110 which proves the first inequality . in order to prove the second inequality ,
let @xmath111 be a fixed primal point such that @xmath112 .
then , we note that the nonnegative function @xmath113 has lipschitz gradient with constant @xmath114 and thus we have @xcite : @xmath115 taking now @xmath116 and using , then we obtain : @xmath117 furthermore , combining with and we have : @xmath118 using the relation @xmath119 we have : @xmath120 which shows the second inequality of our lemma .
this lemma will play a major role in proving rate of convergence for the methods presented in this paper .
note that in @xmath94 enters linearly , while in @xcite @xmath94 enters quadratically in the context of augmented lagrangian and thus in the sequel we will get better convergence estimates than those in the previous papers . in conclusion , for solving the dual problem in duquad we use the following inexact ( augmented ) dual first order algorithm : recall that @xmath121 satisfying the inner criterion and @xmath122 .
moreover , @xmath61 is chosen as follows : * * dgm * : in ( augmented ) dual gradient method @xmath62 , where @xmath63 for all @xmath64 , or equivalently @xmath65 for all @xmath64 , i.e. the ordinary gradient algorithm . * * dfgm * : in ( augmented ) dual fast gradient method @xmath68 , where @xmath123 and @xmath70 , i.e. a variant of nesterov s accelerated scheme .
therefore , in duquad we can solve the smooth ( augmented ) dual problem either with dual gradient method * dgm * ( @xmath124 ) or with dual fast gradient method * dfgm * ( @xmath61 is updated based on @xmath125 ) .
recall that for computing @xmath126 in duquad we use algorithm * fom*(@xmath127 ) ( see the discussion of step 1 ) . when applied to inner subproblem , algorithm * fom*(@xmath127 ) will converge linearly provided that @xmath128 .
moreover , when applying algorithm * fom*(@xmath127 ) we use warm start : i.e. we start our iteration from previous computed @xmath129 . combining the inexact descent relation with lemma [ lemma_sublin_dfg ]
we obtain the following convergence rate for the general algorithm * dfom*(@xmath130 ) in terms of dual function values of : @xcite [ lemma_sublin_dfg_inexact ] for the smooth ( augmented ) dual problem the dual sequences @xmath131 generated by algorithm * dfom*(@xmath130 ) satisfy the following convergence estimate on dual suboptimality : @xmath132 where recall @xmath133 and @xmath81 is defined as in .
note that in ( * ? ? ?
* theorem 2 ) , the convergence rate of * dgm * scheme is provided in the average dual iterate @xmath134 and not in the last dual iterate @xmath135 .
however , for a uniform treatment in theorem [ lemma_sublin_dfg_inexact ] we redefine the dual final point ( the dual last iterate @xmath135 when some stopping criterion is satisfied ) as follows : @xmath136_{{\mathcal{k}}_d}$ ] .
we now show how to choose the inner accuracy @xmath94 in duquad . from theorem [ lemma_sublin_dfg_inexact ]
we conclude that in order to get @xmath86-dual suboptimality , i.e. @xmath137 , the inner accuracy @xmath94 and the number of outer iteration @xmath138 ( i.e. number of updates of lagrange multipliers ) have to be chosen as follows : @xmath139 indeed , by enforcing each term of the right hand side of to be smaller than @xmath140 we obtain first the bound on the number of the outer iterations @xmath138 . by replacing this bound into the expression of @xmath94
, we also obtain how to choose @xmath94 , i.e the estimates .
we conclude that the inner qp has to be solved with higher accuracy in dual fast gradient algorithm * dfgm * than in dual gradient algorithm * dgm*. this shows that dual gradient algorithm * dgm * is robust to inexact ( augmented ) dual first order information , while dual fast gradient algorithm * dfgm * is sensitive to inexact computations ( see also fig .
[ dfgm_sensitivity ] ) . in duquad
the user can choose either algorithm * dfgm * or algorithm * dgm * for solving the ( augmented ) dual problem and he can also choose the inner accuracy @xmath94 for solving the inner problem ( in the toolbox the default values for @xmath94 are taken of the same order as in ) .
for a strongly convex qp with @xmath141.,title="fig:",scaledwidth=52.0%,height=170 ] for a strongly convex qp with @xmath141.,title="fig:",scaledwidth=52.0%,height=170 ]
it is natural to investigate how to recover an @xmath86-primal solution for the original qp . since dual suboptimality
is given in the last dual iterate @xmath135 , it is natural to consider as an approximate primal solution the last primal iterate generated by algorithm * dfom * in @xmath135 , i.e. : @xmath142 note that the last primal iterate @xmath143 coincides with @xmath121 for algorithm * dgm*. however , for algorithm * dfgm * these two sequences are different , i.e. @xmath144
. we will show below that the last primal iterate @xmath145 is an @xmath146-primal solution for the original qp , provided that @xmath137 .
we can also construct an approximate primal solution based on an average of all previous primal iterates generated by algorithm * dfom * , i.e. : @xmath147 recall that @xmath148 in algorithm * dgm * and @xmath149 is updated according to the rule @xmath150 and @xmath70 in algorithm * dfgm*. in the sequel , we prove that the average of primal iterates sequence @xmath151 is an @xmath86-primal solution for the original qp , provided that @xmath137 . before proving primal rate of convergence for algorithm * dfom
* we derive a bound on @xmath152 , with @xmath135 generated by algorithm * dfom * , bound that will be used in the proofs of our convergence results . in the case of * dgm * , using its particular iteration , for any @xmath153 , we have : @xmath154 taking now @xmath155 and using an inductive argument , we get : @xmath156 on the other hand , for the scheme * dfgm * , we introduce the notation @xmath157 and present an auxiliary result : @xcite [ th_tseng_2 ] let @xmath158 be generated by algorithm * dfom*(@xmath159 ) with @xmath123 , then for any lagrange multiplier @xmath46 we have : @xmath160 for all @xmath161 , where @xmath162 . using this result and a similar reasoning as in @xcite
we obtain the same relation for the scheme * dfgm*. moreover , for simplicity , in the sequel we also assume @xmath163 . in the next two sections we derive rate of convergence results of algorithm *
dfom * in both primal sequences , the primal last iterate and an average of primal iterates . in this section
we present rate of convergence results for the algorithm * dfom * , in terms of primal suboptimality and infeasibility for the last primal iterate @xmath164 defined in , provided that the relations hold .
let @xmath165 be some desired accuracy and @xmath166 be the primal last iterate sequence generated by algorithm * dfom*(@xmath167 ) using the inner accuracy from . then , after @xmath138 number of outer iterations given in
, @xmath168 is @xmath146-primal optimal for the original qp . using the lipschitz property of the gradient of @xmath169
, it is known that the following inequality holds @xcite : @xmath170 taking @xmath171 and using the optimality condition @xmath172 for all @xmath173 , we further have : @xmath174 considering @xmath175 and observing that @xmath176 we obtain a link between the primal feasibility and dual suboptimality : @xmath177 provided that @xmath178 and using @xmath94 as in , we obtain : @xmath179 secondly , we find a link between the primal and dual suboptimality .
indeed , we have for all @xmath180 : @xmath181_{{\mathcal{k } } } \right \rangle.\end{aligned}\ ] ] further , using the cauchy - schwartz inequality , we derive : @xmath182 on the other hand , from the concavity of @xmath169 we obtain : @xmath183 taking @xmath184 and @xmath94 as in , based on and on the implicit assumption that @xmath185 , we observe that @xmath186 for both schemes * dgm * and * dfgm*. therefore , implies : @xmath187 as a conclusion , from and the previous inequality , we get the bound : @xmath188 which implies @xmath189 . using this fact and
the feasibility bound , which also implies @xmath190 , we finally conclude that the last primal iterate @xmath191 is @xmath146-primal optimal .
we can also prove linear convergence for algorithm * dfom * provided that @xmath192 ( i.e. the objective function is smooth and strongly convex ) and @xmath193 ( i.e. the inner problem is unconstrained ) . in this case
we can show that the dual problem satisfies an error bound property @xcite .
under these settings * dfom * is converging linearly ( see @xcite for more details ) .
further , we analyze the convergence of the algorithmic framework * dfom * in the average of primal iterates @xmath195 defined in .
since we consider different primal average iterates for the schemes * dgm * and * dfgm * , we analyze separately the convergence of these methods in @xmath195 .
let @xmath165 be some desired accuracy and @xmath194 be the primal average iterate given in , generated by algorithm * dgm * , i.e. algorithm * dfom*(@xmath167 ) with @xmath196 for all @xmath161 , using the inner accuracy from . then , after @xmath138 number of outer iterations given in
, @xmath197 is @xmath86-primal optimal for the original qp .
first , we derive sublinear estimates for primal infeasibility for the average primal sequence @xmath195 ( recall that in this case @xmath198 ) .
given the definition of @xmath199 in algorithm * dfom*(@xmath200 ) with @xmath201 , we get : @xmath202_{\mathcal{k}_d } \quad \forall j \geq 0.\ ] ] subtracting @xmath203 from both sides , adding up the above inequality for @xmath204 to @xmath205 , we obtain : @xmath206_{\mathcal{k}_d } -\lambda^j \right\| = \frac{1}{k+1}{\lvert\lambda^{k+1 } - \lambda^0\rvert } .\end{aligned}\ ] ] if we denote @xmath207_{\mathcal{k}_d } $ ] , then we observe that @xmath208 .
thus , we have @xmath209 . using the definition of @xmath210
, we obtain : @xmath211 using @xmath212 and the bound for the values @xmath94 and @xmath184 from in the previous inequality , we get : @xmath213 it remains to estimate the primal suboptimality .
first , to bound below @xmath214 we proceed as follows : @xmath215_{\mathcal{k } } \rangle \nonumber\\ & \le f(\hat{u}^k_{\epsilon } ) + { \lvert\lambda^*\rvert } { \lvertg\hat{u}^k_{\epsilon}+g - \left[g\hat{u}^k_{\epsilon } + g\right]_{\mathcal{k}}\rvert}\nonumber\\ & = f(\hat{u}^k_{\epsilon } ) + r_d \ ; \text{dist}_{\mathcal{k } } \left(g\hat{u}^k_{\epsilon } + g\right ) .
\end{aligned}\ ] ] combining the last inequality with , we obtain : @xmath216 secondly , we observe the following facts : for any @xmath217 , @xmath218 and the following identity holds : @xmath219 based on previous discussion , and , we derive that @xmath220 taking now @xmath221 , @xmath184 and using an inductive argument , we obtain : @xmath222 provided that @xmath163 . from , and
, we obtain that the average primal iterate @xmath197 is @xmath86-primal optimal .
further , we analyze the primal convergence rate of algorithm * dfgm * in the average primal iterate @xmath194 : let @xmath165 be some desired accuracy and @xmath194 be the primal average iterate given in , generated by algorithm * dfgm * , i.e. algorithm * dfom*(@xmath167 ) with @xmath223 for all @xmath224 , using the inner accuracy from . then , after @xmath138 number of outer iterations given in , @xmath197 is @xmath86-primal optimal for the original qp .
recall that we have defined @xmath225 .
then , it follows : @xmath226 for any @xmath227 we denote @xmath228 and thus we have @xmath229_{\mathcal{k}_d}$ ] . in these settings
, we have the following relations : @xmath230_{\mathcal{k}_d})\right ) \nonumber\\ & = \theta_j \left(\left [ \mu^{j } + \frac{1}{2l_{\text{d } } } \bar{\nabla } d_{\rho}(\mu^{j } ) \right]_{\mathcal{k}_d } - \lambda^{j}\right ) \nonumber\\ & = \theta_{j}(\lambda^{j } - \mu^{j})\nonumber\\ & = \theta_{j}(\lambda^{j } - \lambda^{j-1 } ) + ( \theta_{j-1 } -1 ) ( \lambda^{j-2 } - \lambda^{j-1})\nonumber\\ & = \underbrace{\lambda^{j-1 } + \theta_{j}(\lambda^{j } - \lambda^{j-1})}_{=l^{j } } - \underbrace{(\lambda^{j-2 } + \theta_{j-1}(\lambda^{j-1 } - \lambda^{j-2}))}_{=l^{j-1}}.\end{aligned}\ ] ] for simplicity consider @xmath231 and @xmath232 . adding up the above equality for @xmath233 to @xmath234 , multiplying by @xmath235 and observing that @xmath236_{\mathcal{k}_d } \in \mathcal{k}$ ] for all @xmath237 , we obtain : @xmath238_{\mathcal{k}_d } ) \right)\right\| \\ & = \left\| \sum\limits_{j=0}^k \frac{\theta_j}{s_k } \left ( \bar{\nabla } d_{\rho}(\mu^j ) - 2 l_{\text{d}}(z^j-[z^j]_{\mathcal{k}_d } ) \right)\right\| \\ & \overset{\eqref{feasibility_aux1}}{= } \frac{l_\text{d}}{s_k}{\lvertl^{k}-l^0\rvert } \le \frac{4l_\text{d}}{(k+1)^2 } { \lvertl^k -l^{0}\rvert}.\end{aligned}\ ] ] taking @xmath239 in lemma [ th_tseng_2 ] and using that the two terms @xmath240 and @xmath241 are positive , we get : @xmath242 thus , we can further bound the primal infeasibility as follows : @xmath243 therefore , using @xmath138 and @xmath94 from , it can be derived that : @xmath244 further , we derive sublinear estimates for primal suboptimality .
first , note the following relations : @xmath245 summing on the history and using the convexity of @xmath246 , we get : @xmath247 using in lemma [ th_tseng_2 ] , and dropping the term @xmath248 , we have : @xmath249 moreover , we have that : @xmath250 now , by choosing the lagrange multiplier @xmath251 and @xmath184 in , we have : @xmath252 on the other hand , we have : @xmath253_{\mathcal{k } } \rangle\nonumber\\ & \le f(\hat{u}^k_{\epsilon } ) + r_d \ ; \text{dist}_{{\mathcal{k}}}(g \hat{u}^k_{\epsilon } + g).\end{aligned}\ ] ] taking @xmath254 and @xmath94 from , and using , we obtain : @xmath255 finally , from , and , we get that the primal average sequence @xmath197 is @xmath86 primal optimal . in conclusion , in duquad we generate two approximate primal solutions @xmath145 and @xmath151 for each algorithm * dgm * and * dfgm*. from previous discussion it can be seen that theoretically , the average of primal iterates sequence @xmath151 has a better behavior than the last iterate sequence @xmath145 . on the other hand , from our practical experience
( see also section [ numerical_tests ] ) we have observed that usually dual first order methods are converging faster in the primal last iterate than in a primal average sequence .
moreover , from our unified analysis we can conclude that for both approaches , ordinary dual with @xmath256 and augmented dual with @xmath11 , the rates of convergence of algorithm * dfom * are the same .
in this section we derive the total computational complexity of the algorithmic framework * dfom*. without lose of generality , we make the assumptions : @xmath257 however , if any of these assumptions does not hold , then our result are still valid with minor changes in constants .
now , we are ready to derive the total number of iterations for * dfom * , i.e. the total number of projections on the set @xmath93 and of matrix - vector multiplications @xmath258 and @xmath259 . [ in : th_last ] let @xmath165 be some desired accuracy and the inner accuracy @xmath94 and the number of outer iterations @xmath138 be as in . by setting @xmath260 and assuming that the primal iterate @xmath126 is obtained by running the algorithm * fom*(@xmath261 ) , then @xmath164 ( @xmath194 ) is @xmath146 ( @xmath86 ) primal optimal after a total number of projections on the set @xmath93 and of matrix - vector multiplications @xmath258 and @xmath259 given by : @xmath262 from lemma [ lemma_sublin_dfg ] we have that the inner problem ( i.e. finding the primal iterate @xmath126 ) for a given @xmath263 can be solved in sublinear ( linear ) time using algorithm * fom*(@xmath264 ) , provided that the inner problem has smooth ( strongly ) convex objective function , i.e. @xmath265 has @xmath266 ( @xmath267 ) .
more precisely , from lemma [ lemma_sublin_dfg ] , it follows that , regardless if we apply algorithms * dfgm * or * dgm * , we need to perform the following number of inner iterations for finding the primal iterate @xmath126 for a given @xmath263 : @xmath268 combining these estimates with the expressions for the inner accuracy @xmath94 , we obtain , in the first case @xmath269 , the following inner complexity estimates : @xmath270 multiplying @xmath271 with the number of outer iterations @xmath138 from and minimizing the product @xmath272 over the smoothing parameter @xmath273 ( recall that @xmath100 and @xmath274 ) , we obtain the following optimal computational complexity estimate ( number of projections on the set @xmath93 and evaluations of @xmath258 and @xmath259 ) : @xmath275 which is attained for the optimal parameter choice : @xmath276 using the same reasoning for the second case when @xmath277 , we observe that the value @xmath278 is also optimal for this case in the following sense : the difference between the estimates obtained with the exact optimal @xmath273 and the value @xmath279 are only minor changes in constants .
therefore , when @xmath277 , the total computational complexity ( number of projections on the set @xmath93 and evaluations of @xmath258 and @xmath259 ) is : @xmath280 in conclusion , the last primal iterate @xmath164 is @xmath146-primal optimal after @xmath281 ( @xmath282 ) total number of projections on the set @xmath93 and of matrix - vector multiplications @xmath258 and @xmath259 , provided that @xmath269 ( @xmath277 ) .
similarly , the average of primal iterate @xmath194 is @xmath86-primal optimal after @xmath281 ( @xmath282 ) total number of projections on the set @xmath93 and of matrix - vector multiplications @xmath258 and @xmath259 , provided that @xmath269 ( @xmath277 ) .
moreover , the optimal choice for the parameter @xmath273 is of order @xmath283 , provided that @xmath38 .
let us analyze now the computational cost per inner and outer iteration for algorithm * dfom*(@xmath284 ) for solving approximately the original qp : * inner iteration * : when we solve the inner problem with the nesterov s algorithm * fom*(@xmath285 ) , the main computational effort is done in computing the gradient of the augmented lagrangian @xmath286 defined in , which e.g. has the form : @xmath287 in duquad these matrix - vector operations are implemented efficiently in c ( matrices that do not change along iterations are computed once and only @xmath288 is computed at each outer iteration ) .
the cost for computing @xmath289 for general qps is @xmath290 .
however , when the matrices @xmath35 and @xmath291 are sparse ( e.g. network utility maximization problem ) the cost @xmath290 can be reduced substantially .
the other operations in algorithm * fom*(@xmath285 ) are just vector operations and thus they are of order @xmath292 .
thus , the dominant operation at the inner stage is the matrix - vector product .
* outer iteration * : when solving the outer ( dual ) problem with algorithm * dfom*(@xmath284 ) , the main computational effort is done in computing the inexact gradient of the dual function : @xmath293 the cost for computing @xmath294 for general qps is @xmath295 .
however , when the matrix @xmath291 is sparse , this cost can be reduced .
the other operations in algorithm * dfom*(@xmath130 ) are of order @xmath296 .
thus the dominant operation at the outer stage is also the matrix - vector product .
[ fig : gprof_n150_dfgm_case1 ] displays the result of profiling the code with gprof . in this simulation , a standard qp with inequality constraints and dimensions @xmath297 and @xmath298
was solved by algorithm * dfgm*. the profiling summary is listed in the order of the time spent in each file .
this figure shows that almost all the time for executing the program is spent in the library module _ math - functions.c_. furthermore , _ mtx - vec - mul _ is by far the dominating function in this list .
this function is multiplying a matrix with a vector , which is defined as a special type of matrix multiplication . in conclusion ,
in duquad the main operations are the matrix - vector products .
therefore , duquad is adequate for solving qp problems on hardware with limited resources and capabilities , since it does not require any solver for linear systems or other complicating operations , while most of the existing solvers for qps from the literature implementing e.g. active set or interior point methods require the capability of solving linear systems . on the other hand
, duquad can be also used for solving large - scale sparse qp problems since the iterations are very cheap in this case ( only sparse matrix - vector products ) .
duquad is mainly intended for small to medium size , dense qp problems , but it is of course also possible to use duquad to solve ( sparse ) qp instances of large dimension .
the duquad software package is available for download from : + + and distributed under general public license to allow linking against proprietary codes .
proceed to the menu point `` software '' to obtain a zipped archive of the most current version of duquad . the users manual and extensive source code documentation are available here as well .
+ an overview of the workflow in duquad is illustrated in fig .
[ fig : duquad_workflow ] .
a qp problem is constructed using a matlab script called _
test.m_. then , the function _
duquad.m _ is called with the problem as input and is regarded as a prepossessing stage for the online optimization .
the binary mex file is called , with the original problem and the extra info as input .
main.c _ file of the c - code includes the mex framework and is able to convert the matlab data into c format .
furthermore , the converted data gets bundled into a c struct and passed as input to the algorithm that solves the problem .
we plot in fig .
[ fig : qp_cpu ] the average cpu time for several solvers , obtained by solving @xmath299 random qp s with equality constraints ( @xmath11 and @xmath19 ) for each dimension @xmath300 , with an accuracy @xmath141 and the stopping criteria @xmath301 and @xmath302 less than the accuracy @xmath86 . in both algorithms * dgm * and * dfgm * we consider the average of iterates @xmath303 .
since @xmath304 , we have chosen @xmath305 . in the case of algorithm * dgm * , at each outer iteration the inner problem is solved with accuracy @xmath306 . for the algorithm * dfgm
* we consider two scenarios : in the first one , the inner problem is solved with accuracy @xmath307 , while in the second one we use the theoretic inner accuracy .
we observe a good behavior of algorithm * dfgm * , comparable to cplex and gurobi .
we plot in fig .
[ fig : comparison_dfo ] the number of iterations of algorithms * dgm * and * dfgm * in the primal last and average iterates for @xmath308 random qps with inequality constraints ( @xmath34 and @xmath18 ) of variable dimension ranging from @xmath309 to @xmath310 .
we choose the accuracy @xmath141 and the stopping criteria was @xmath311 and @xmath312 less than the accuracy @xmath86 . from this figure
we observe that the number of iterations are not varying much for different test cases and also that the number of iterations are mildly dependent on problem s dimension .
finally , we observe that dual first order methods perform usually better in the primal last iterate than in the average of primal iterates .
-0.1 cm , @xmath18 ) for * dgm * and * dfgm * in primal last / average of iterates for different test cases of the same dimension ( left ) and of variable dimension ( right ) .
, title="fig:",scaledwidth=105.0%,height=188 ] a. domahidi , a. zgraggen , m. zeilinger , m. morari and c. jones , _ efficient interior point methods for multistage problems arising in receding horizon control _ , ieee conference decision and control , 668674 , 2012 .
j. jerez , k. ling , g. constantinides and e. kerrigan , _ model predictive control for deeply pipelined field - programmable gate array implementation : algorithms and circuitry _ , iet control theory and applications , 6(8):10291041 , 2012 .
v. nedelcu , i. necoara and q. tran dinh , _ computational complexity of inexact gradient augmented lagrangian methods : application to constrained mpc _ , siam journal control and optimization , 52(5):31093134 , 2014 .
i. necoara , l. ferranti and t. keviczky , _ an adaptive constraint tightening approach to linear mpc based on approximation algorithms for optimization _ , optimal control applications and methods , doi : 10.1002/oca.2121 , 1 - 19 , 2015 .
s. richter , m. morari and c. n. jones , _ towards computational complexity certification for constrained mpc based on lagrange relaxation and the fast gradient method _ , ieee conference decision and control , 52235229 , 2011 . r.d .
zimmerman , c.e .
murillo - sanchez and r.j .
thomas , _ matpower : steady - state operations , planning , and analysis tools for power systems research and education _ , ieee transactions on power systems , 26(1):1219 , 2011 . | in this paper we present the solver duquad specialized for solving general convex quadratic problems arising in many engineering applications .
when it is difficult to project on the primal feasible set , we use the ( augmented ) lagrangian relaxation to handle the complicated constraints and then , we apply dual first order algorithms based on inexact dual gradient information for solving the corresponding dual problem .
the iteration complexity analysis is based on two types of approximate primal solutions : the primal last iterate and an average of primal iterates .
we provide computational complexity estimates on the primal suboptimality and feasibility violation of the generated approximate primal solutions .
then , these algorithms are implemented in the programming language c in duquad , and optimized for low iteration complexity and low memory footprint .
duquad has a dynamic matlab interface which make the process of testing , comparing , and analyzing the algorithms simple .
the algorithms are implemented using only basic arithmetic and logical operations and are suitable to run on low cost hardware .
it is shown that if an approximate solution is sufficient for a given application , there exists problems where some of the implemented algorithms obtain the solution faster than state - of - the - art commercial solvers . |
one of the factors of the enormous success of our societies is our ability to cooperate @xcite . while in most animal species cooperation
is observed only among kin or in very small groups , where future interactions are likely , cooperation among people goes far beyond the five rules of cooperation @xcite : recent experiments have shown that people cooperate also in one - shot anonymous interactions @xcite and even in large groups @xcite .
this poses an evolutionary puzzle : why are people willing to pay costs to help strangers when no future rewards seem to be at stake ?
a growing body of experimental research suggests that cooperative decision - making in one - shot interactions is most likely a history - dependent dynamic process . _
dynamic _ because time pressure @xcite , cognitive load @xcite , conceptual priming of intuition @xcite , and disruption of the right lateral prefrontal cortex @xcite have all been shown to promote cooperation , providing direct evidence that automatic actions are , on average , more cooperative than deliberate actions .
_ history - dependent _ because it has been found that previous experience with economic games on cooperation and intuition interact such that experienced subjects are less cooperative than inexperienced subjects , but only under time pressure @xcite and that intuition promotes cooperative behavior only among inexperienced subjects with above median trust in the setting where they live @xcite .
while this latter paper also shows that promoting intuition versus reflection has no effect among experienced subjects , its results are inconclusive with regard to people with little trust in their environment , due to the limited number of observations . more generally , the limitation of previous studies is that they have all been conducted in developed countries and so they do not allow to draw any conclusions about what happens among people from a societal background in which they are exposed to frequent non - cooperative acts .
two fundamental questions remain then unsolved .
what is the effect of promoting intuition versus deliberation among people living in a non - cooperative setting ?
how does this interact with previous experience with economic games on cooperative decision - making ?
the first question is particularly intriguing since , based on existing theories , several alternatives are possible .
the social heuristics hypothesis ( shh ) , introduced by rand and colleagues @xcite to explain the intuitive predisposition towards cooperation described above `` posits that cooperative decision making is guided by heuristic strategies that have generally been successful in one s previous social interactions and have , over time , become internalized and automatically applied to social interactions that resemble situations one has encountered in the past .
when one encounters a new or atypical social situation that is unlike previous experience , one generally tends to rely on these heuristics as an intuitive default response .
however , through additional deliberation about the details of the situation , one can override this heuristic response and arrive at a response that is more tailored to the current interaction '' @xcite .
then , according to the shh , inexperienced subjects living in a non - cooperative setting should bring their non - cooperative strategy ( learned in the setting where they live ) in the lab as a default strategy .
these subjects are then predicted to act non - cooperatively both under time pressure , because they use their non - cooperative default strategy , and under time delay , because defection is optimal in one - shot interactions . however , this is not the only possibility .
several studies have shown that patients who suffered ventromedial prefrontal cortex damage , which causes the loss of emotional responsiveness , are more likely to display anti - social behavior @xcite .
these findings support the interpretation that intuitive emotions play an important role in pro - social behavior and form the basis of haidt s social intuition model ( sim ) according to which moral judgment is caused by quick moral intuitions and is followed ( when needed ) by slow , ex post facto , moral reasoning @xcite . while the sim does not make any prediction on what happens in the specific domain of cooperation , it would certainly be consistent with a general intuitive predisposition towards cooperation , mediated by positive emotions , and independent of the social setting in which an individual is embedded .
a third alternative is yet possible .
motivated by work suggesting that people whose self - control resources have been taxed tend to cheat more @xcite and be less altruistic @xcite , it has been argued that self - control plays an important role in overriding selfish impulses and bringing behavior in line with moral standards .
this is consistent with kohlberg s rationalist approach @xcite , which assumes that moral choices are guided by reason and cognition : as their cognitive capabilities increase , people learn how to take the other s perspective , which is fundamental for pro - social behavior .
this rationalist approach makes the explicit prediction that promoting intuition always undermine cooperation . in sum
, the question of how promoting intuition versus reflection affects cooperative behavior among people living in a non - cooperative setting is far from being trivial and , based on existing theories , all three possibilities ( positive effect , negative effect , no effect ) are , a priori , possible . concerning previous experience on economic games , while the sim and the rationalist approach do not make any prediction about its role on cooperative decision - making among people living in a non - cooperative setting
, the shh predicts that it has either a null or a positive effect driven by intuitive responses . this because experienced participants , despite their living in a non - cooperative setting , _ might _ have internalized a cooperative strategy to be used only in experiments . of course , the shh does not predict that a substantial proportion of subjects have _ in fact _ developed such a context - dependent cooperative intuition - and this is why the predicted effect is _ either _ positive _ or _ null . in the former case
, however , the shh predicts that the positive effect should be driven by intuitive responses , since the shh assumes that experience operates primarily through the channel of intuition .
here we report on an experiment aimed at clarifying these points .
we provide evidence of two major results : ( i ) promoting intuition versus reflection has no effect on cooperation among subjects living in a non - cooperative setting and with no previous experience with economic games on cooperation ; ( ii ) experienced subjects are more cooperative than inexperienced subjects , but only when acting under time pressure . taken together
, these results suggest that cooperation is a learning process , rather than an instinctive impulse or a self - controlled choice , and that experience operates primarily via the channel of intuition .
in doing so , they shed further light on human cooperative decision - making and provide further support for the social heuristics hypothesis .
we have conducted an experiment using the online labor market amazon mechanical turk ( amt ) @xcite recruiting participants only from india .
india is a particularly suited country to hire people from for our purpose : if , as many studies have confirmed @xcite , good institutions are crucial for the evolution of cooperation , and if , as many scholars have argued @xcite , corruption and cronyism are endemic in indian society , then residents in india are likely to have very little trust on strangers and so they are likely to have internalized non - cooperative strategies in their every - day life .
one study confirms this hypothesis , by showing that spiteful preferences are widespread in the village of uttar pradesh and this ultimately implies residents inability to cooperate @xcite . at the same time , according to demographic studies on amt population @xcite , india is the second most active country on amt after the us , which facilitates the procedure of collecting data .
participants were randomly assigned to either of two conditions : in the _ time pressure _ condition we measured intuitive cooperation ; in the _ time delay _ condition we measured deliberate cooperation . as a measure of cooperation
, we adopted a standard two - person prisoner s dilemma ( pd ) with a continuous set of strategies .
specifically , participants were given an endowment of @xmath2 , and asked to decide how much , if any , to transfer to the other participant .
the amount transferred would be multiplied by 2 and earned by the other participant ; the remainder would be earned by themselves , but without being multiplied by any factor .
each participant was informed that the other participant was facing the same decision problem .
participants in the time pressure condition were asked to make a decision within 10 seconds and those in the time delay condition were asked to wait for at least 30 seconds before making their choice . after making their decision , participants had to answer four comprehension questions , after which they entered the demographic questionnaire , where , along with the usual questions , we also asked `` to what extent have you previously participated in other studies like to this one ( e.g. , exchanging money with strangers ) ? '' using a 5 point likert - scale from `` never '' to `` several times '' .
as in previous studies @xcite , we used the answer to this question as a measure of participant s previous experience with economic games on cooperative decision - making .
as in these studies , we say that a subject is _ inexperienced _ if he or she answered `` never '' to the above question .
in the supplementary online material we also report the results of a pilot , in which we measured participants level of experience by asking them to report the extent to which they had participated in _
exactly _ the same task before .
although the use of the word `` exactly '' may lead to confusion , with some minor differences in the details , our main results are robust to the use of this measure ( see supplementary online material for more details ) . after collecting the results ,
bonuses were computed and paid on top of the participation fee ( @xmath3 ) . no deception was used .
a total of 949 subjects participated in our experiment . taken globally ,
results contain a lot of noise , since only 449 subjects passed the comprehension questions . here
we restrict our analysis to subjects who passed all comprehension questions and we refer the reader to the supplementary online material for the analysis of those subjects who failed the attention check .
we include in our analysis also subjects who did not obey the time constraint in order to avoid selection problems that impair causal inference @xcite .
first we ascertain that our time manipulation effectively worked . analyzing participants decision times , we find that those in the time delay condition ( @xmath4 ) took , on average , 45.64 seconds to make their decision , while those under time pressure took , on average , only 20.04 seconds .
thus , although many subjects under time pressure did not obey the time constraint , the time manipulation still had a substantial effect .
participants under time pressure transferred , on average , 27.93% of their endowment , and those under time delay transferred , on average , 28.57% of their endowment .
linear regression using time manipulation as a dummy variable confirms that the difference is not statistically significant ( coeff @xmath5 , @xmath6 ) , even after controlling for age , sex , and level of education ( coeff @xmath7 @xmath8 ) .
we note that restricting the analysis to subjects who obeyed the time constraint leads to qualitatively equivalent results ( 26.11% under time pressure vs 29.02% under time delay , coeff @xmath9 , @xmath10 ) .
thus , promoting intuition versus reflection does not have any effect on cooperative decision - making . _ en passant _ , we note that these percentages are far below those observed among us residents in a very similar experiment @xcite .
more precisely , in this latter paper , us residents started out with a @xmath11 endowment and were asked to decide how much , if any , to give to the other person .
as in the current study , the amount transferred would be multiplied by 2 and earned by the other player .
strictly speaking , these two experiments are not comparable for three reasons .
first , in @xcite there was no time manipulation ; second , the initial endowments were different ; third , stakes used in the experiment in the us did not correspond to the same stakes in indian currency . according to previous research ,
these differences are minor .
indeed , recent studies have argued that stakes do not matter as long as they are not too high @xcite and that neutrally framed pds give rise to a percentage of cooperation sitting between that obtained in the time pressure condition and that obtained in the time delay condition @xcite . thus comparing the percentage of cooperation in the current study ( 28% ) with that reported in @xcite ( 52% ) supports our assumption that the average indian sample is particularly non - cooperative or , at least , less cooperative than the average us sample .
next we investigate our main research questions .
figure 1 summarizes our results , providing visual evidence that ( i ) promoting intuition versus reflection has no significant effect on cooperation among inexperienced subjects ; that ( ii ) experienced subjects cooperate more than inexperienced subjects , but only when acting under time pressure . more specifically , we find that inexperienced subjects under time pressure ( @xmath12 ) transferred , on average , @xmath13 of their endowment while those under time delay ( @xmath14 ) transferred , on average , @xmath15 of their endowment .
the difference is not significant ( coeff @xmath16 , @xmath17 ) , even after controlling for all socio - demographic variables ( coeff @xmath18 , @xmath19 ) .
thus , promoting intuition versus reflection has no effect on cooperation among inexperienced subjects living in a non - cooperative setting .
this finding is robust to controlling for people who did not obey the time constraint ( coeff @xmath20 , @xmath21 ) and to controlling for people who obeyed the time contraint ( coeff @xmath22 , @xmath23 ) . to explore the interaction between experience and cooperative behavior , as in previous studies
@xcite , we separate subjects into experienced and inexperienced .
this procedure comes from the observation that , although the level of experience is a categorical variable , the association between participant s objective level of experience and their answer to our question is objective only in case of inexperienced subjects .
linear regression predicting cooperation using experience as a dummy variable confirms that experience with economic games on cooperation favors the emergence of cooperative choices , but only among people in the time pressure condition ( time pressure : coeff @xmath24 , @xmath25 ; time delay : coeff @xmath26 , @xmath27 ) .
these results are robust to including control on the socio - demographic variables ( time pressure : coeff @xmath28 , @xmath29 ; time delay : coeff @xmath30 , @xmath31 ) .
our main results are also robust to using non - parametric tests , such as wilcoxon rank - sum : the rate of cooperation of inexperienced subjects is not statistically distinguishable from the rate of cooperation of inexperienced subjects acting under time delay both when they act under time pressure ( @xmath32 ) and under time delay ( @xmath33 ) ; and the rate of cooperation of inexperienced subjects is significantly smaller than that of experienced subjects , but only among those acting under time pressure ( time pressure : @xmath34 ; time delay : @xmath35 ) . for completeness , we also report the results of linear regression predicting cooperation using level of experience as independent variable .
we find that level of experience has a marginally significant positive effect on cooperation among subjects acting under time pressure ( coeff @xmath36 , @xmath37 ) and has no effect on cooperation among subjects acting under time delay ( coeff = @xmath38 , @xmath39 ) .
also these results are robust to including control on all the socio - demographic variables ( time pressure : coeff @xmath40 , @xmath41 ; time delay : coeff @xmath42 , @xmath43 ) .
the increase of cooperation from inexperienced subjects to experienced subjects seems to be driven by participants under time pressure who did _ not _ obey the time constraint .
specifically , linear regression predicting cooperation using experience as a dummy variable yields non - significant results in case of participants who obeyed the time pressure condition ( without control : coeff @xmath44 , @xmath45 ; with control : coeff @xmath46 , @xmath47 ) and significant results in case of participants who did not obey it ( without control : coeff @xmath48 , @xmath49 ; with control : coeff @xmath50 , @xmath51 ) .
this is not surprising and it is most probably due to noise generated by a combination of two factors : the set of people who obeyed the time pressure contraint is very small ( for instance , only 14 inexperienced people obeyed the time constraint ) and it is more likely to contain people who did not understand the decision problem but passed the comprehension questions by chance ( which we estimated to be 5% of the total .
see supplementary online material ) . ) .
previous experience with economic games on cooperation has a positive effect on cooperation , but only among participants in the time pressure condition ( @xmath25 ) . _ _ ]
we have shown that ( i ) promoting intuition via time pressure versus promoting deliberation via time delay has no effect on cooperative behavior among subjects residents in india with no previous experience with economic games on cooperation , and that ( ii ) experience has a positive effect on cooperation , but this effect is significant only among subjects acting under time pressure .
our results have several major implications , the first of which is providing further support for the social heuristics hypothesis ( shh ) @xcite .
introduced in order to organize the growing body of literature providing direct @xcite and indirect @xcite evidence that , on average , intuitive responses are more cooperative than reflective responses , the shh contends that people internalize strategies that are successful in their everyday social interactions and then apply them to social interactions that resemble situations they have encountered in the past .
thus , when they encounter a new or atypical situation , people tend to rely on these heuristics and use them as intuitive responses .
deliberation can override these heuristics and adjust the behavior towards one that is more tailored to the current interaction .
as such , the shh makes a prediction that has not been tested so far : inexperienced subjects living in a non - cooperative setting should act non - cooperatively both under time pressure , because they use their non - cooperative default strategy ( learned in the setting where they live ) , and under time delay , because defection is optimal in one - shot interactions .
our results support this prediction . besides this prediction , the shh is also consistent with an interaction between level of previous experience with economic games on cooperation , time pressure , and cooperation in one - shot interactions : experienced people , despite their living in a non - cooperative setting , _ might _ have internalized a cooperative strategy , to be used only in amt .
the shh does not predict that a substantial proportion of experienced people have _ in fact _
developed this context - dependent intuition for cooperation , but it is certainly consistent with a positive effect of experience on cooperation driven by intuitive responses .
our results provide evidence for this phenomenon .
as mentioned in the introduction , kohlberg s rationalist approach makes the explicit prediction that promoting intuition should always undermine cooperation .
thus our results support the shh versus kohlberg s rationalistic approach .
of course , this does _ not _ imply that the rationalist approach should be completely rejected : it is indeed supported by many experimental studies involving pro - social behaviors other than cooperation . if anything , our results point out that different pro - social behaviors may emerge from different cognitive processes .
classifying pro - social behaviors in terms of the processes involved is an important direction for future research towards which , to the best of our knowledge , only one recent study has attempted a first step @xcite . supporting the shh
, our results suggest that economic models of human cooperation should start taking dual processes and individual history into account .
indeed , virtually all major models of human cooperation are static and decontextualized and only a handful of papers have recently attempted a first step in the direction of taking dual processes into account @xcite . we believe that extending these approaches to incorporate also individual history could be a promising direction for future research .
our findings go beyond the mere support of the shh . our cross cultural analysis , although it is formally not correct , shows that residents in india are , on average , less cooperative than us residents .
the difference is so large ( 28% vs 52% ) that it is hard to explain it by appealing to minor differences in the experimental designs and so it deserves to be commented .
one possibility , supported by the experimental evidence that good institutions are crucial in promoting cooperation @xcite and the evidence that india struggles on a daily basis to fight corruption in politics at both the national and local levels @xcite , is that residents in india may have internalized non cooperative behavior in their everyday life ( because cooperation is not promoted by their institutions ) and they tend to apply it also to the new situation of a lab experiment .
one far - reaching consequence of this interpretation is that the role of local institutions may go far beyond regularizing behavior .
if institutions do not support cooperative behavior , selfishness may even get internalized and applied to atypical situations where people rely on heuristics .
while this interpretation is supported by a recent study @xcite showing that norms of cooperation learned in one experiment spill over to subsequent experiments where there are no norms , we recommend caution on our interpretation , since our results do _ not _ show directly that inexperienced residents in india are less cooperative than us residents _ because _ they are embedded into a society whose institutions do not promote cooperative behavior .
however , we believe that this is a fundamental point that deserves to be rigorously addressed in further research .
interestingly , we have shown that experienced residents in india are significantly more cooperative than inexperienced ones .
this correlation appears to be even more surprising if seen in light of recent studies reporting that experience has a _ negative _ effect on cooperation among residents in the us @xcite .
although the sign of these effects are different , they share the property that they are driven by intuitive responses .
thus they are in line with the shh , which assumes that experience operates mainly through the channel of intuition , but it does not make any prediction about the sign of the effect of experience , which may ultimately depend on a number of factors . while it is relatively easy to explain a negative effect of experience with economic games on cooperation , by appealing to learning of the payoff maximizing strategy , explaining a positive effect is harder .
one possibility is that experienced subjects have learned cooperation in iterated games , where it might be strategically advantageous , and tend to apply it also in one - shot games .
another possibility is that turkers are developing a feeling of community that may favor the emergence of pro - social preferences .
understanding what mechanisms can promote the emergence of cooperation from a non - cooperative setting is certainly a fundamental topic for further research .
1 trivers rl .
1971 the evolution of reciprocal altruism .
* 46 * , 35 - 57 .
axelrod r , hamilton wd .
1981 the evolution of cooperation .
_ science _ * 211 * , 1390 - 1396 .
ostrom e. 2000 collective action and the evolution of social norms .
_ j. econ .
perspect . _
* 14 * , 137 - 158 .
milinski m , semmann d , kranbeck hj .
2002 reputation helps solve the ` tragedy of the commons ' .
_ nature _ * 415 * , 424 - 426 .
fehr e , fischbacher u. 2003 the nature of human altruism .
_ nature _ * 425 * , 785 - 791 .
boyd r , gintis h , bowles s , richerson pj .
2003 the evolution of altruistic punishment .
sci usa _ * 100 * , 3531 - 3535 .
perc m , szolnoki a. 2010 coevolutionary games : a mini review .
_ biosystems _ * 99 * , 109 - 125 .
apicella cl , marlowe fw , fowler jh , christakis na . 2012 social networks and cooper- ation in hunter - gatherers .
_ nature _ * 481 * , 497 - 501 .
capraro v. 2013 a model of human cooperation in social dilemmas .
_ plos one _ * 8 * , e72427 .
crockett mj .
2013 models of morality . _ trends .
* 17 * , 363 - 366 .
rand dg , nowak ma .
2013 human cooperation .
_ trends cogn .
_ * 17 * , 413 - 425 .
zaki j , mitchell jp .
2013 intuitive prosociality .
* 22 * , 466 - 470 .
capraro v , halpern jy .
2014 translucent players : explaining cooperative behavior in social dilemmas .
_ available at ssrn 2509678_. hauser op , rand dg , peysakhovich a , nowak ma .
2014 cooperating with the future .
_ nature _ * 511 * , 220 - 223 .
bone je , wallace b , bshary r , raihani nj .
2014 the effect of power asymme- tries on cooperation and punishment in a prisoners dilemma game . _
plos one _ , * 10 * , e0117183 .
2006 five rules for the evolution of cooperation .
_ science _ * 314 * , 1560 - 1563 .
wong ry , hong y. 2005 dynamic influences of culture on cooperation in the prisoner s dilemma .
_ psychol .
_ * 16 * , 429 - 434 .
khadjavi m , lange a. 2013 prisoners and their dilemma .
_ j. econ .
_ , * 92 * , 163 - 175 .
capraro v , jordan jj , rand dg .
2014 heuristics guide the implementation of social preferences in one - shot prisoner s dilemma experiments .
* 4 * , 6790 .
capraro v , smyth c , mylona k , niblo ga .
2014 benevolent characteristics promote cooperative behaviour among humans . _
plos one _ * 9 * , e102881 .
capraro v , marcelletti a. 2014 do good actions inspire good actions in others ? _
_ * 4 * , 7470 .
barcelo h , capraro v. 2015 group size effect on cooperation in one - shot social dilemmas .
rep . _ * 5 * , 7937 .
rand dg , greene jd , nowak ma .
2012 spontaneous giving and calculated greed .
_ nature _ * 489 * , 427 - 430 .
rand dg , peysakhovich a , kraft - todd gt , newman ge , wurzbacher o , nowak ma , greene jd .
2014 social heuristics shape intuitive cooperation .
commun . _ * 5 * , 3677 .
cone j , rand dg .
2014 time pressure increases cooperation in competitively framed social dilemmas . _
plos one _ , * 9 * , e115756 .
rand dg , kraft - todd gt .
2014 reflection does not undermine self - interested prosociality .
_ * 8 * , 300 .
rand dg , newman ge , wurzbacher om .
2014 social context and the dynamics of cooperative choice .
_ j. behav .
_ doi:10.1002/bdm.1837 .
schulz jf , fischbacher u , thni c , utikal v. 2012 affect and fairness : dictator games under cognitive load .
_ j. econ .
psychol . _ * 41 * , 77 - 87 .
cornelissen g , dewitte s , warlop l. 2011 are social value orientations expressed automatically ?
decision making in the dictator game .
_ pers . social psychol .
_ doi : 10.1177/0146167211405996 .
roch sg , lane ja , samuelson cd , allison st , dent jl .
2000 cognitive load and the equality heuristic : a two - stage model of resource overconsumption in small groups .
proc . _ * 83 * , 185 - 212 .
lotz s. 2014 spontaneous giving under structural inequality : intuition promotes cooperation in asymmetric social dilemmas . _ available at ssrn : http://ssrn.com / abstract=2513498_. ruff c , ugazio g , fehr e. 2013 changing social norm compliance with noninvasive brain stimulation . _ science _ * 342 * , 482 - 484 .
damasio ar , tranel d , damasio h. 1990 individuals with sociopathic behavior caused by frontal damage fail to respond autonomically to social stimuli . _ behav .
brain res . _ * 41 * , 81 - 94 .
bechara a , damasio ar , damasio h , anderson sw .
1994 insensitivity to future consequences following damage to human prefrontal cortex .
_ cognition _ * 50 * , 7 - 15 .
bechara a , tranel d , damasio h , damasio ar .
1996 failure to respond autonomically to anticipated future outcomes following damage to prefrontal cortex .
cortex _ * 6 * , 215 - 225 .
haidt j. 2001 the emotional dog and its rational tail : a social intuitionist approach to moral judgment .
_ psychol .
rev . _ * 108 * , 814 - 834 .
mead nl , baumeister rf , gino f , schweitzer me , ariely d. 2009 too tired to tell the truth : self - control resource depletion and dishonesty . _
psychol . _ * 45 * , 594 - 597 .
gino f , schweitzer me , mead nl , ariely d. 2011 unable to resist temptation : how self- control depletion promotes unethical behavior .
* 115 * , 191 - 203 .
halali e , bereby - meyer y , ockenfels a. 2013 is it all about the self ?
the effect of self- control depletion on ultimatum game proposers .
neurosci . _ * 7 * , 240 .
xu h , bgue l , bushman bj .
2012 too fatigued to care : ego depletion , guilt , and prosocial behavior .
psychol . _ * 48 * , 1183 - 1186 .
achtziger a , als - ferrer c , wagner ak .
2014 money , depletion , and prosociality in the dictator game .
_ j. neurosci .
_ in press .
kohlberg l. 1963 the development of childrens orientations toward a moral order .
dev . _ * 6 * , 11 - 33 .
paolacci g , chandler j , ipeirotis pg .
2010 running experiments on amazon mechanical turk .
* 5 * , 411 - 419 .
horton jj , rand dg , zeckhauser rj .
2011 the online laboratory : conducting experiments in a real labor market . _
* 14 * , 399 - 425 .
2012 the promise of mechanical turk : how online labor markets can help theorists run behavioral experiments .
_ j. theor .
biol . _ * 299 * , 172 - 179 .
bardhan p. 2000
irrigation and cooperation : an empirical analysis of 48 irriga- tion communities in south india .
. change _ * 48 * , 847 - 865 .
dayton - johnson j. 2000 determinants of collective action on the local commons : a model with evidence from mexico .
* 62 * , 181 - 208 .
fujiie m , hayami y , kikuchi m. 2005 the conditions of collective action for local commons management : the case of irrigation in the philippines .
econ . _ * 33 * , 179 - 189 .
henrich j , boyd r , bowles s , camerer c , fehr e , gintis h , et al .
2005 `` economic man '' in cross - cultural perspective : behavioral experiments in 15 small - scale societies .
brain sci .
_ * 28 * , 795 - 815 .
henrich j , mcelreath r , barr a , ensminger j , barrett c , bolyanatz a , et al .
2006 costly punishment across human societies .
_ science _ * 312 * , 1767 - 1770 .
herrmann b , thni c , gchter s. 2008 antisocial punishment across societies .
_ science _ * 319 * , 1362 - 1367 .
buchan nr , grimalda g , wilson r , brewer m , fatas e , foddy m. 2009 globalization and human cooperation .
usa _ * 106 * , 4138 - 4142 .
gchter s , herrmann b. 2009 reciprocity , culture and human cooperation : previous insights and a new cross - cultural experiment .
b _ * 364 * , 791 - 806 . gchter s , herrmann b , thni c. 2010 culture and cooperation . _
b _ * 365 * , 2651 - 2661 .
buchan nr , brewer mb , grimalda g , wilson rk , fatas e , foddy m. 2011 global social identity and global cooperation .
_ psychol .
* 22 * , 821 - 828 .
andersson k , agrawal a. 2011 inequalities , institutions , and forest commons .
_ global environ . change _ * 21 * , 866 - 875 .
bigoni m , bortolotti s , casari m , gambetta d , pancotto f. 2014 amoral familism , social capital , or trust ? the behavioral foundations of the italian north - south divide . _ working paper_. das sk .
2001 _ public office , private interest : bureaucracy and corruption in india . _ oxford university press .
guha r. 2007 _ india after gandhi : the history of the world s largest democracy_. pan .
2008 curbing corruption in india : an impossible dream ? _ asian j. polit .
* 16 * , 240 - 259 .
miklian j , carney s. 2013 corruption , justice and violence in democratic india .
_ sais rev .
_ * 33 * , 37 - 49 .
fehr e , hoff k , kshetramade m. 2008 spite and development . _
rev . _ * 98 * , 494 - 499 .
ross j , irani l , silberman m , zaldivar a , tomlinson b. 2010 who are the crowdworkers ? : shifting demographics in mechanical turk .
_ chi10 extended abstracts on human factors in computing systems _ , 2863 - 2872 .
tinghg g , andersson d , bonn c , bttiger h , josephson c , lundgren g , et al .
2013 intuition and cooperation reconsidered .
_ nature _ * 498 * , e1-e2 .
amir o , rand dg , gal yk .
2012 economic games on the internet : the effect of $ 1 stakes . _ plos one _ * 7 * , e31461 .
engel c , rand dg .
2014 what does clean really mean ?
the implicit framing of decontextualized experiments .
lett . _ * 122 * , 386 - 389 .
kieslich pj , hilbig be .
2014 cognitive conflict in social dilemmas : an analysis of response dynamics .
mak . _ * 9 * , 510 - 522 .
righetti f , finkenauer c , finkel ej .
2013 low self - control promotes the willingness to sacrifice in close relationships .
_ psychol .
_ doi:10.1177/0956797613475457 .
rand dg , epstein zg .
2014 risking your life without a second thought : intuitive decision - making and extreme altruism . _
plos one _ * 9 * , e109687 .
corgnet b , espn am , hernn - gonzlez r. 2015 the cognitive basis of social behavior : cognitive reflection overrides antisocial but not always prosocial motives . _ available on researchgate_. fudenberg d , levine dk .
2006 a dual - self model of impulse control .
* 96 * , 1449 - 1476 .
fudenberg d , levine dk .
2011 risk , delay , and convex self - control costs . _ am .
j. microecon .
_ * 3 * , 34 - 68 .
fudenberg d , levine dk .
2012 timing and self - control .
_ econometrica _ * 80 * , 1 - 42 .
dreber a , fudenberg d , levine dk , rand dg .
2014altruism and self - control .
_ available at ssrn : http://ssrn.com / abstract=2477454_. peysakhovich a , rand dg . in press .
habits of virtue : creating norms of cooperation and defection in the laboratory .
_ this supplementary information is divided in three sections : in the first section we report more details about the statistical analysis ; in the second section we report full instructions of our experiment . in the third section
we report the results of our pilot ( containing a methodological error in the measure of participants level of experience ) . with some differences in the details ,
these results are in line with those reported in the main text and thus provide further evidence in support of our main result that experienced subjects cooperate more than inexperienced subjects and that experience operates primarily through the channel of intuition .
we start by reporting the socio - demographics of participants who passed the comprehension questions .
data are summarized in the table .
we remind that participants level of experience was measured using a 5-point likert - scale from 1=never to 5=several times .
the socio - demographic statistics show that the majority of subjects in the time pressure condition did not obey the time constraint .
this is likely due to the fact that reading the instructions of the decision problem takes about six seconds and so participants had only 4 seconds to understand the problem and make their decision . however , the mean decision time of participants in the time pressure condition was much smaller than the mean decision time of participants in the time delay condition , providing evidence that the time manipulation still had a substantial effect . [ cols="<,^,^",options="header " , ]
finally we analyze subjects who failed the attention test . indeed , the large number of participants who failed the comprehension questions ( about half of the total number ) is worrisome that a substantial fraction of those who passed the comprehension questions may have passed them just by chance .
although this rate of passing is not much lower than that in similar studies conducted in the us ( in a very similar experiment , conducted with american subjects and published in @xcite , 32% of subjects did not pass the comprehension questions ) it could potentially generate noise in our data .
indeed , subjects who failed the attention test played essentially at random ( time pressure : average transfer = 47.20% ; time delay : average transfer = 46.35% ) . to exclude this possibility ,
we analyse failers responses to the comprehension questions .
we start with the first comprehension question , which asked `` what is the choice by you that maximizes your outcome ? '' .
participants could choose any even amount of money from 0c to 20c , for a total of 11 possible choices .
figure 1 reports the distribution of responses of people who failed at least one comprehension questions .
the distribution is clearly tri - modal , with 60% of responses equally distributed among the two extreme choices ( full cooperation and full defection ) and the mid - point ( transfer half ) .
the remaining 40% is equally distributed among all other choices .
consequently , assuming that confused participants respond to the first comprehension question according to this distribution , the probability that a confused participant pass the first comprehension question by chance is equal to 1/5 .
next we analyze the responses to the second comprehension question , which asked `` what is the choice by you that maximizes the other participant s outcome ? '' .
figure 2 reports the distribution of answers .
also in this case we find a tri - modal distribution , although this time the correct answer appeared with higher frequency ( about 50% ) . assuming that a confused participant answered this question according to this distribution , we conclude that a confused participant had probability about 1/10 to pass the first two comprehension questions by chance .
then we analyze the responses to the third comprehension question , which asked `` what is the choice by the other participant that maximizes your outcome ? ''
figure 3 reports the distribution of answers .
this time the distribution is essentially bi - modal , with about 3/5 of people answering correctly . assuming that confused participants answered according to this distribution , we conclude that a confused participant had probability about 3/50 to pass the first three comprehension questions by chance .
now , it is impossible to have a clue of what the proportion of confused participants who passed the fourth comprehension question by chance is .
the analysis above suggests that 3/50 is an upper bound of the probability that a confusing subject passed all comprehension questions by chance .
interestingly , 88% of subjects who answered 20c in the third comprehension and failed the fourth comprehension question , answered 20c also to the fourth comprehension question , which asked `` what is the choice by the other participant that maximizes the other participant s outcome ? '' .
this suggests that the actual probability of passing all comprehension questions by chance is much lower that 3/50 . in any case , although our results do not allow to make a precise estimation of noise , the proportion of people who passed the comprehension questions by chance is likely below 5% of the total , suggesting that noise is a minor problem in our data .
participants were randomly assigned to either the time pressure condition or the time delay condition . in both conditions , after entering their worker i d , participants were informed that they would be asked to make a choice in a decision problem to be presented later and that comprehension questions would be asked .
participants were also informed that the survey ( which was made using the software qualtrics ) contained a skip logic which would automatically exclude all participants failing any of the comprehension questions . specifically , this screen was as follows : _ welcome to this hit .
_ _ this hit will take about five minutes .
for the participation to this hit , you will earn 0.50 us dollars , that is , about 31 inr .
you can also earn additional money depending on the decisions that you and the other participants will make .
_ you will be asked to make one decision .
there is no incorrect answer .
however : _ _ important : after making the decision , to make sure you understood the decision problem , we will ask some simple questions , each of which has only one correct answer . if you fail to correctly answer any of those questions , the survey will automatically end and you will not receive any redemption code and consequently you will not get any payment .
_ _ with this in mind , do you wish to continue ? _ at this stage , they could either leave the study or continue .
those who decided to continue were redirected to an introductory screen where we gave them all the necessary information about the decision problem , but without telling exactly which one it is .
this is important in order to have the time pressure and time delay conditions work properly in the next screen .
this introductory screen for the participants in the time pressure condition was the following : _ you have been paired with another participant .
you can earn additional money depending on the decision you will make in the next screen .
you will be asked to make a choice that can affect your and the other participant s outcome .
the decision problem is symmetric : also the other participant is facing the same decision problem .
after the survey is completed , you will be paid according to your and the other participant s choices .
_ _ you will have only 10 seconds to make the choice .
_ _ this is the only interaction you have with the other participant .
he or she will not have the opportunity to influence your gain in later parts of the hit .
if you are ready , go to the next page .
_ the introductory screen for the participants in the time delay condition was identical , a part from the fact that the sentence ` you will have only 10 seconds to make the choice ' was replaced by the sentence ` you will be asked to think for at least 30 seconds before making your choice .
use this time to think carefully about the decision problem ' .
the decision screen was the same in both conditions : _ you and the other participant are both given @xmath2 us dollars .
you and the other participant can transfer , independently , money to the each other .
every cent you transfer , will be multiplied by @xmath52 and earned by the other participant .
every cent you do not transfer , will be earned by you .
_ _ how much do you want to transfer ? _ by using appropriate buttons , participants could transfer any even amount of money from @xmath53 to @xmath2 .
en passant , we observe that reading the decision screen takes about six seconds and thus participants under time pressure had only about four seconds to make their choice . to assure that time pressure and time delay work properly ,
it is necessary that comprehension questions are asked after the decision has been made .
thus , right after the decision screen , participants faced the following four comprehension questions . _ what is the choice by you that maximizes your outcome ? _
_ what is the choice by you that maximizes the other participant s outcome ? _ _ what is the choice by the other participant that maximizes your outcome ?
_ _ what is the choice by the other participant that maximizes the other participant s outcome ? _ by using appropriate buttons , participants could select any even amount of money from @xmath53 to @xmath2 .
participants who failed any of the comprehension questions were automatically excluded from the survey . those who answered all questions correctly entered the demographic questionnaire , where we asked for their age , sex , reason for their choice , and , most importantly , level of experience in these games . specifically , we asked the following question : _ to what extent have you previously participated in other studies like to this one ( e.g. , exchanging money with strangers ) ? _ answers
were collected using a 5 point likert - scale from `` 1=never '' to `` 5=several times '' .
our pilot experiment was identical to our main experiment , except for the fact that , as a measure of experience , we asked participants to self - report the extent to which they had participated in `` exactly '' the same task before
. participants could choose between : never , once or twice , and several times .
the use of the word `` exactly '' is problematic , since it might lead to confusion : what does the answer `` exactly the same task '' imply ?
does it imply that participants have participated in a task with exactly the same instructions ( including time constrains ) or does it imply that the participant have participated in a task containing the same economic game ?
figure 4 reports the results of the pilot .
we observe that , indeed , the details are different : level of experience seem to have a inverted - u effect on cooperation , which , since it has not been replicated in the main experiment , is probably due to confusion regarding the interpretation of the word `` exactly '' .
however , as in our main experiment , we find that experienced subjects are significantly more cooperative than little or no experienced subjects and that this behavioral change is mainly driven by intuitive responses ( see figure 5 ) .
specifically , linear regression confirms that little experienced subjects are significantly less cooperative than inexperienced subjects both under time pressure ( coeff @xmath54 , @xmath55 ) and under time delay ( coeff @xmath56 , @xmath57 ) ; and confirms that experienced subjects are significantly more cooperative than little experienced subjects both under time pressure ( coeff @xmath58 , @xmath59 ) and under time delay ( coeff @xmath60 , @xmath61 ) .
moreover , experienced subjects were also significantly more cooperative than inexperienced subjects , both under time pressure ( coeff @xmath62 , p @xmath63 ) and under time delay ( coeff @xmath64 , p @xmath63 ) .
thus experience has a significant inverted - u effect on cooperation , where little experienced subjects cooperate the least and experienced subjects the most .
the coefficients of the previous regressions suggest that the motivations behind the initial decrease of cooperation , which affects subjects under time pressure and those under time delay to exactly the same extent , are different from the motivations behind the subsequent flourishing of cooperation , which seems to affect subjects under time pressure to a larger extent than those under time delay . to confirm this , we use linear regression to predict decision among experienced subjects using time pressure as a dummy variable .
we find that experienced subjects under time pressure are nearly significantly more cooperative than experienced subjects under time delay ( coeff @xmath65 , @xmath66 ) . | recent studies suggest that cooperative decision - making in one - shot interactions is a history - dependent dynamic process : promoting intuition versus deliberation has typically a positive effect on cooperation ( dynamism ) among people living in a cooperative setting and with no previous experience in economic games on cooperation ( history - dependence ) . here
we report on a lab experiment exploring how these findings transfer to a non - cooperative setting .
we find two major results : ( i ) promoting intuition versus deliberation has no effect on cooperative behavior among inexperienced subjects living in a non - cooperative setting ; ( ii ) experienced subjects cooperate more than inexperienced subjects , but only under time pressure .
these results suggest that cooperation is a learning process , rather than an instinctive impulse or a self - controlled choice , and that experience operates primarily via the channel of intuition . in doing so ,
our findings shed further light on the cognitive basis of human cooperative decision - making and provide further support for the recently proposed social heuristics hypothesis .
@xmath0center for mathematics and computer science ( cwi ) , 1098 xg , amsterdam , the netherlands .
@xmath1department of political sciences , luiss guido carli , 00197 , roma , italy .
contact author : [email protected]__forthcoming in proceedings of the royal society b : biological sciences _ _ |
in this paper , we consider the problem of the nonparametric density deconvolution of @xmath0 , the density of identically distributed variables @xmath1 , from a sample @xmath7 in the model @xmath8 where the @xmath1 s and @xmath9 s are independent sequences , the @xmath9 s are i.i.d . centered random variables with common density @xmath10 , that is @xmath11 is a noise with known density @xmath12 and known noise level @xmath13 . due to the independence between the @xmath1 s and the @xmath9 s ,
the problem is to estimate @xmath0 using the observations @xmath14 with common density @xmath15 the function @xmath16 is often called the convolution kernel and is completely known here . denoting by @xmath17 the fourier transform of @xmath18 ,
it is well known that since @xmath19 , two factors determine the estimation accuracy in the standard density deconvolution problem : the smoothness of the density to be estimated , and the one of the error density which are described by the rate of decay of their fourier transforms . in this context ,
two classes of errors are usually considered : first the so called ordinary smooth " errors with polynomial decay of their fourier transform and second , the super smooth " errors with fourier transform having an exponential decay . for further references about density deconvolution see e.g. carroll and hall ( 1988 ) , devroye ( 1989 ) , fan ( 1991a , b ) , liu and taylor ( 1989 ) , masry ( 1991 , 1993a , b ) , stefansky ( 1990 ) , stefansky and carroll ( 1990 ) , taylor and zhang ( 1990 ) , zhang ( 1990 ) and cator ( 2001 ) , pensky and vidakovic ( 1999 ) , pensky ( 2002 ) , fan and koo ( 2002 ) , butucea ( 2004 ) , butucea and tsybakov ( 2004 ) , koo ( 1999 ) . the aim of the present paper is to provide a complete simulation study of the deconvolution estimator constructed by a penalized contrast minimization on a model @xmath20 , a space of square integrable functions having a fourier transform with compact support included into @xmath21 $ ] with @xmath22 .
comte et al .
( 2005 ) show that for @xmath23 being a positive integer , this penalized contrast minimization selects the relevant projection space @xmath20 without any prior information on the unknown density @xmath0 . in most cases
, it is an adaptive estimator in the sense that it achieves the optimal rate of convergence in the minimax sense , studied by fan ( 1991a ) , butucea ( 2004 ) and butucea and tsybakov ( 2004 ) .
it is noteworthy that , contrary to what usually happens , @xmath24 does not correspond here to the dimension of the projection space but to the length of the support of the fourier transform of the functions of @xmath25 .
thus we will refer in the following to @xmath24 as the `` length '' of the model @xmath25 . moreover , in the context of integer @xmath23 , comte et al .
( 2005 ) provide a brief simulation which shows that the selected @xmath23 are rather small and therefore far from the asymptotic .
our present study shows that it is relevant to choose @xmath22 on a thinner grid than one included in @xmath26 .
thus we start by stating a modification of the results in comte et al .
( 2005 ) to take into account this thinner grid of values @xmath24 and we show that the resulting penalized minimum contrast estimator is an adaptive estimator in the sense that it achieves the optimal rate of convergence in the minimax sense . here , the penalty depends on the smoothness of the errors density and therefore we consider two cases : laplace density ( ordinary smooth ) and gaussian density ( super smooth ) .
we illustrate , through examples , the influence of over - penalization and under - penalization and propose practical calibrations of the penalty in all considered cases
. then we study in very large simulations the non asymptotic properties of our estimator by considering various types of densities @xmath0 , with various smoothness properties like cauchy distribution , gaussian density and finally fjer - de - la - valle poussin - type density .
we present some examples , that illustrate how the algorithm works .
we give the mean integrated squared error ( mise ) for the two types of errors density , for all the test densities , for various @xmath13 , and for various sample size .
our results present global tables of mise and comparisons between mise and the theoretical expected rates of convergence .
lastly , the robustness of our procedure is tested in various ways : when the observations are dependent , when @xmath13 is very small ( leading to a problem of density estimation ) and when the errors density @xmath27 is misspecified or not taken into account . in those cases ,
we compare our procedure with previous results of delaigle and gjibels ( 2004a , 2004b ) and dalelane ( 2004 ) ( direct density estimation ) .
the conclusions of our study are the following .
our estimation procedure provides very good results ; better than the kernel deconvolution methods described and studied in delaigle and gijbels ( 2004a ) .
our estimation procedure is robust when the @xmath28 s are no longer independent and even not strongly mixing .
we underline the importance of the noise level in the quality of estimation , and we check that , in the case of a very small noise , we obtain mise s that have the same order as some recent results obtained by dalelane ( 2004 ) for direct density estimation .
lastly our results show that a misspecification of the errors density slightly increases the error of estimation , but less than the use of the direct density estimator ( without deconvolving ) , as it was already mentioned in hesse ( 1999 ) .
> from a practical point of view it is important to note that our algorithm is a fast algorithm ( @xmath29 operations ) based on the inverse fast fourier transform ( ifft ) . the paper is organized as follows . in section 2 , we present the model , the assumptions , the adaptive estimator and its expected rates of convergence . in section 3 , we describe the implementation of the estimates ( see [ descri ] ) and the computations of the associated integrated squared errors ( [ mise ] ) .
section 4 presents the chosen penalties ( see [ penalites ] ) and describes the framework of our simulations .
the simulation results are gathered in section 5 and an appendix is devoted to the proof of our theorem .
for @xmath18 and @xmath30 two square integrable functions , we denote by @xmath17 the fourier transform of @xmath18 , @xmath31 and by @xmath32 the convolution product , @xmath33 .
moreover , we denote by @xmath34 .
consider model ( [ model ] ) under the following assumptions .
@xmath35 under assumption ( [ iid ] ) , the @xmath28 s are independent and identically distributed random variables . assumption ( [ fepspair ] ) , usual for the construction of an estimator in density deconvolution
, ensures that @xmath0 is identifiable . the rate of convergence for estimating @xmath0 is strongly related to the rate of decrease of the fourier transform of the errors density @xmath36 as @xmath37 goes to infinity . more precisely , the smoother @xmath27 , the quicker the rate of decay of @xmath38 and the slower the rate of convergence for estimating @xmath0 .
indeed , if @xmath27 is very smooth , so is @xmath39 the density of the observations @xmath40 and thus it is difficult to recover @xmath0 .
this decrease of @xmath41 is described by the following assumption .
@xmath42 when @xmath43 in assumption ( [ regufeps ] ) , @xmath27 is usually called `` ordinary smooth '' , and when @xmath44 and @xmath45 , the error density is usually called `` super smooth '' . indeed densities satisfying assumption ( [ regufeps ] ) with @xmath45 and @xmath44
are infinitely differentiable .
for instance , gaussian or cauchy distributions are super smooth of order @xmath46 and @xmath47 respectively , and the symmetric exponential ( also called laplace ) distribution with @xmath48 and @xmath49 is an ordinary smooth density .
furthermore , when @xmath50 , ( [ fepspair ] ) requires that @xmath51 in ( [ regufeps ] ) . by convention , we set @xmath52 when @xmath43 and we assume that @xmath44 when @xmath53 . in the same way , if @xmath54 , the @xmath1 s are directly observed without noise and we set @xmath55 . for the construction of the estimator
we need the following more technical assumption .
@xmath56 this assumption ( [ momg ] ) , quite unusual but unrestrictive , already appears in density deconvolution in a slightly different way in pensky and vidakovic ( 1999 ) who assume , instead of ( [ momg ] ) that @xmath57 .
the main drawback of this condition is that it is not stable by translation , but an empirical centering of the data seems to avoid practical problems . since rates of convergence depend on the smoothness of @xmath0 we introduce regularity conditions .
@xmath58}(x)\right\}.\notag\end{aligned}\ ] ] note that densities satisfying ( [ super ] ) with @xmath59 belong to some sobolev class of order @xmath60 , whereas densities satisfying ( [ super ] ) with @xmath61 are infinitely differentiable .
moreover , such densities admit analytic continuation on a finite width strip when @xmath62 and on the whole complex plane if @xmath63 .
the densities satisfying ( [ entiere ] ) , often called entire functions , admit analytic continuation on the whole complex plane ( see ibragimov and hasminskii ( 1983 ) ) . in order to clarify the notations , we denote by greek letters the parameters related to the known distribution of the noise @xmath64 and by latin letters the parameters related to the unknown distribution @xmath0 of @xmath65 .
let @xmath66 and @xmath67 .
using that @xmath68 is an orthonormal basis of the space of square integrable functions having a fourier transform with compact support included into @xmath69=[-\ell_m , \ell_m]$ ] ( see meyer ( 1990 ) ) , we denote by @xmath25 such a space and consider the collection of linear spaces @xmath70 , with @xmath71 , @xmath72 , and @xmath73 with @xmath74 , as projection spaces .
consequently , @xmath75\},\end{aligned}\ ] ] and the orthogonal projection of @xmath0 on @xmath25 , @xmath76 is given by @xmath77 , with @xmath78 . since
this orthogonal projection involves infinite sums , we consider in practice , the truncated spaces @xmath79 defined as @xmath80 where @xmath81 is an integer to be chosen later .
associated to those spaces we consider the orthogonal projection of @xmath0 on @xmath79 denoted by @xmath82 and given by @xmath83 associate this collection of models to the following contrast function , for @xmath84 belonging to some @xmath25 of the collection @xmath85 @xmath86 since @xmath87 = \langle t , g\rangle,$ ] we find that @xmath88 which is minimum when @xmath89 .
since @xmath90 estimates the @xmath91 distance between @xmath84 and @xmath0 , it is well adapted for estimating @xmath0 .
associated to the collection of models , the collection of the non penalized estimators @xmath92 is defined by @xmath93 by using that @xmath94 is linear , and that @xmath95 is an orthonormal basis of @xmath79 , we have @xmath96 with @xmath97 the adaptive estimator is computed by using the following penalized criteria @xmath98,\ ] ] where pen ( . ) is a penalty function based on the observations and the known distribution of @xmath99 without any prior information on @xmath0 .
first , the variance term @xmath102 depends , as usual in deconvolution problems , on the rate of decay of the fourier transform of @xmath27 , with larger variance for smoother @xmath27 . under assumption ( [ regufeps ] ) , for @xmath103 , the variance term satisfies @xmath104 where @xmath105 second , under assumption ( [ momg ] ) , @xmath106 is of order @xmath107 .
consequently , under ( [ regufeps ] ) , @xmath108 ensures that the risk @xmath109 has the order @xmath110 finally , the bias term @xmath111 depends on the smoothness of the function @xmath0 and has the expected order for classical smoothness classes since it is given by the distance between @xmath0 and the classes of entire functions having fourier transform compactly supported on @xmath112 $ ] ( see ibragimov and hasminskii ( 1983 ) ) .
consequently , under ( [ momg ] ) , if @xmath108 , the rate of convergence of @xmath92 is obtained by selecting the space @xmath79 , and thus @xmath117 , that minimizes @xmath118 one can see that if @xmath24 becomes too large , the risk explodes , due to the presence of the second term .
hence @xmath24 appears to be the cut between the relevant low frequencies used in the fourier transforms to compute the estimate and the high frequencies which are not used ( and may even degrade the quality of the risk ) .
we give the resulting rates in table [ rates ] . for a density
@xmath0 satisfying ( [ super ] ) , rates are , in most cases , known to be the optimal one in the minimax sense ( see fan ( 1991a ) , butucea ( 2004 ) , butucea and tsybakov ( 2004 ) ) .
we refer to comte et al .
( 2005 ) for further discussion about optimality . | we consider the problem of estimating the density @xmath0 of identically distributed variables @xmath1 , from a sample @xmath2 where @xmath3 , @xmath4 and @xmath5 is a noise independent of @xmath1 with known density @xmath6 .
we generalize adaptive estimators , constructed by a model selection procedure , described in comte et al .
( 2005 ) .
we study numerically their properties in various contexts and we test their robustness .
comparisons are made with respect to deconvolution kernel estimators , misspecification of errors , dependency , ...
it appears that our estimation algorithm , based on a fast procedure , performs very well in all contexts .
universit paris v , map5 , umr cnrs 8145 . ] iut de paris v et universit dorsay , laboratoire de probabilits , statistique et modlisation , umr 8628 . ]
adaptive estimation
. density deconvolution .
model selection .
penalized contrast .
projection estimator . simulation study .
data - driven . |
wigner defined a quantum elementary particle as a system whose hilbert space of states carries an irreducible representation of the poincar group @xcite .
this definition is a group theoretical one .
this lead the author to look for a definition of a classical elementary particle by group theoretical methods , relating its definition to the kinematical group structure .
a classical elementary particle was defined as a lagrangian system whose kinematical space is a homogeneous space of the poincar group @xcite .
when quantising these classical models it is shown that the wave function of the system transforms with a projective unitary irreducible representation of the kinematical group @xcite .
the different classical models of spinning particles produced by this formalism are collected in the book @xcite .
one of the models , which will be considered in this work , satisfies dirac s equation when quantised .
the latest lep experiments at cern suggest that the electron charge is confined within a region of radius @xmath0 m .
nevertheless , the quantum mechanical behaviour of the electron appears at distances of the order of its compton s wave length @xmath1 m , which is six orders of magnitude larger .
one possibility to reconcile these features in order to obtain a model of the electron is the assumption from the classical viewpoint that the charge of the electron is just a point but at the same time this point is never at rest and it is affected by the so called _ zitterbewegung _ and therefore it is moving in a confined region of size @xmath2 .
this is the basic structure of the spinning particle models obtained within the kinematical formalism developed by the author @xcite-@xcite and also suggested by dirac s analysis of the internal motion of the electron @xcite .
there , the charge of the particle is at a point @xmath3 , but this point is not the center of mass of the particle . in general
we obtain that the point charge satisfies a fourth order differential equation which , as we shall see , is the most general differential equation satisfied by any three - dimensional curve .
the charge is moving around the center of mass in a kind of harmonic or central motion .
it is this motion of the charge that gives rise to the spin and dipole structure of the particle . in particular , the classical model that when quantised satisfies dirac s equation shows , for the center of mass observer , a charge moving at the speed of light in circles of radius @xmath4 and contained in a plane orthogonal to the spin direction @xcite .
it is this classical model of electron we shall consider in the subsequent analysis and which is reviewed in section [ sec : elec ] . in this article we shall find the group invariant dynamical equations of these classical systems .
the difference of the approach presented here with the previous published works is that the dynamical equations are obtained by group theoretical arguments without any appeal to a lagrangian formalism while there they were obtained by lagrangian methods .
neverteless the dynamical equations obtained are the same .
the article is organised as follows : section 2 is a reminder that the most general differential equation satisfied by a curve in three - dimensional space is of a fourth order . in section 3
we introduce the classical model of a spinning electron that has been shown to satisfy dirac s equation when quantised .
section 4 states the general method for obtaining the group invariant differential equation satisfied by a point and for any arbitrary kinematical group .
this method is applied in sections 5 to 7 to the galilei and poincar groups to obtain the non - relativistic and relativistic invariant dynamical equations of a spinless particle and of the spinning model .
finally , as an application of the obtained dynamical equations , the analysis of the coulomb interaction between two spinning electrons is considered in section 8 .
one of the salient features is the classical prediction of the possible existence of bound states for spinning electron - electron interaction . if two electrons have their center of masses separated by a distance greater than compton s wave length they always repel each other .
but if two electrons have their center of masses separated by a distance less than compton s wave length they can form bound states provided some conditions on their relative spin orientation and center of mass position and velocity are fulfilled .
let us remind that for an arbitrary three - dimensional curve @xmath5 when expressed in parametric form in terms of the arc length @xmath6 , the three orthogonal unit vectors @xmath7 , @xmath8 called respectively tangent , normal and binormal , satisfy the so called frenet - serret differential equations : @xmath9 where @xmath10 and @xmath11 are respectively the curvature and torsion .
since the unit tangent vector is @xmath12 , taking succesive derivatives it yields @xmath13 then elimination of the @xmath7 between these equations implies that the most general curve in three - dimensional space satisfies the fourth - order differential equation : @xmath14 all the coefficients in front of the derivatives @xmath15 can be expressed in terms of the scalar products @xmath16 , @xmath17 .
let us mention that for helical motions there is a constant relationship @xmath18 constant , and therefore the coefficient of @xmath19 vanishes .
we present here the main features of a spinning electron model obtained through the general lagrangian formalism .
the poincar group can be parameterised in terms of the variables @xmath20 , where @xmath21 is dimensionless and represents the relative orientation of inertial frames and the others have dimensions of time , length and velocity , respectively , and represent the corresponding parameters for time and space translation and the relative velocity between observers .
their range is @xmath22 , @xmath23 , @xmath24 but constrained to @xmath25 and @xmath26 .
one of the important homogeneous spaces of the poincar group that defines the kinematical space of a classical elementary particle @xcite is spanned by the variables @xmath20 with the same range as before but now @xmath27 restricted to @xmath28 . as variables of a kinematical space of a classical elementary system they are interpreted as the time , position , velocity and orientation observables of the particle , respectively .
the lagrangian associated to this system will be a funtion of these variables and their next order time derivative .
therefore the lagrangian will also depend on the acceleration and angular velocity .
it turns out that as far as the position @xmath3 is concerned dynamical equations will be of fourth order .
these are the equations we want to obtain in this work but from a different and non - lagrangian method . when analysed the invariance of any lagrangian defined on this manifold under the poincar group , in particular under pure lorentz transformations and rotations , we get respectively the following noether constants of the motion which have the form @xmath29 , or its essential components @xmath30 and @xmath31 in three - vector notation @xcite : @xmath32 @xmath33 observables @xmath34 and @xmath35 are the constant energy and linear momentum of the particle , respectively .
the linear momentum is not lying along the velocity @xmath27 , and @xmath36 is the spin of the system .
it satisfies the dynamical equation @xmath37 which is the classical equivalent of the dynamical equation satisfied by the dirac spin observable . for the center of mass observer
it is a constant of the motion .
this observer is defined by the requirements @xmath38 and @xmath39 .
the first condition implies that the center of mass is at rest and the second that is located at the origin of the frame .
the energy of the particle in this frame is @xmath40 , so that from ( [ eq : k ] ) we get , @xmath41 it turns out that point @xmath3 , which does not represent the center of mass position , describes the motion depicted in figure [ fig1 ] .
it is the position of the charge .
it is orthogonal to the constant spin and also to the velocity as it corresponds to a circular motion at the velocity @xmath42 .
the radius and angular velocity of the internal classical motion of the charge are , respectively , @xmath43 , and @xmath44 .
if we take the time derivative of the constant @xmath45 of ( [ eq : k ] ) and the scalar product with @xmath27 we get @xmath46 where we have a linear relationship between @xmath34 and @xmath35 .
dirac equation is just the quantum mechanical expression of this poincar invariant formula @xcite .
the spin @xmath47 , has two parts : one @xmath48 related to the orbital motion of the charge and another @xmath49 due to the rotation of the particle and which is directly related to the angular velocity as it corresponds to a spherically symmetric object .
the positive energy particle has the total spin @xmath36 oriented in the same direction as the @xmath48 part , as shown in the figure .
the orientation of the spin is the opposite for the negative energy particle , which corresponds to the time reversed motion . when quantizing the system , the orbital component @xmath48 which is directly related to the magnetic moment , quantizes with integer values while the rotational part @xmath49 with half integer values , so that for spin @xmath50 particles the total spin is half the value of the @xmath51 part .
when expressing the magnetic moment in terms of the total spin we get in this way a pure kinematical interpretation of the @xmath52 gyromagnetic ratio @xcite . for the center of mass observer
this system looks like a system of three degrees of freedom .
two represent the @xmath53 and @xmath54 coordinates of the point and the third is the phase of its rotational motion .
however this phase is exactly equal to the phase of the orbital motion of the charge and because the motion is of constant radius at the constant speed @xmath42 then only remains one independent degree of freedom .
therefore the system is equivalent to a one - dimensional harmonic oscillator of frequency @xmath55 .
when quantizing the system the stationary states of the harmonic oscillator have the energy @xmath56 but if the system is elementary then it has no excited states and in the c.m .
frame it is reduced to the ground state of energy @xmath57 so that when compared with the classical result @xmath58 , implies that the constant classical parameter @xmath59 is necessarily @xmath60 .
the radius of the internal charge motion is half compton s wave length .
it is this classical model of electron we shall analyse in subsequent sections , and our interest is to obtain the dynamical equation satisfied by the point charge @xmath3 for any arbitrary relativistic and nonrelativistic inertial observer . to end this section and with the above model in mind
let us collect the main results obtained by dirac when he analysed the motion of the free electron @xcite .
let point @xmath61 be the position vector on which dirac s spinor @xmath62 is defined .
when computing the velocity of point @xmath61 , dirac arrives at : 1 .
the velocity @xmath63=c{\mbox{\boldmath$\alpha$}}$ ] , is expressed in terms of @xmath21 matrices and writes , _
@xmath64 a measurement of a component of the velocity of a free electron is certain to lead to the result @xmath65_. this conclusion is easily seen to hold also when there is a field present. 2 .
the linear momentum does not have the direction of this velocity @xmath66 , but must be related to some average value of it : @xmath67 _ ` the @xmath68 component of the velocity , @xmath69 , consists of two parts , a constant part @xmath70 , connected with the momentum by the classical relativistic formula , and an oscillatory part , whose frequency is at least @xmath71 , @xmath67'_. 3 . about the position @xmath61 : _
` the oscillatory part of @xmath68 is small , @xmath67 , which is of order of magnitude @xmath72 , @xmath67'_. and when analysing , in his original 1928 paper @xcite the interaction of the electron with an external electromagnetic field , after performing the square of dirac s operator , he obtains two new interaction terms : @xmath73 where the electron spin is written as @xmath74 and @xmath75 in terms of @xmath76-pauli matrices and @xmath77 and @xmath78 are the external electric and magnetic fields , respectively .
he says , _ `
the electron will therefore behave as though it has a magnetic moment @xmath79 and an electric moment @xmath80 .
the magnetic moment is just that assumed in the spinning electron model ' _
( pauli model ) . _ the electric moment , being a pure imaginary , we should not expect to appear in the model .
it is doubtful whether the electric moment has any physical meaning. _ in the last sentence it is difficult to understand why dirac , who did not reject the negative energy solutions and therefore its consideration as the antiparticle states , disliked the existence of this electric dipole which was obtained from his formalism on an equal footing as the magnetic dipole term . in quantum electrodynamics , even in high energy processes ,
the complete dirac hamiltonian contains both terms , perhaps in a rather involved way because the above expression is a first order expansion in the external fields considered as classical commuting fields . properly speaking this electric dipole
does not represent the existence of a particular positive and negative charge distribution for the electron . in the classical model , the negative charge of
the electron is at a single point but because this point is not the center of mass , there exists a nonvanishing electric dipole moment with respect to the center of mass of value @xmath81 in the center of mass frame .
its correspondance with the quantum dirac electric moment is shown in @xcite .
i think this is the observable dirac disliked .
it is oscillating at very high frequency and it basically plays no role in low energy electron interactions because its average value vanishes , but it is important in high energy or in very close electron - electron interactions .
let us consider the trajectory @xmath82 , @xmath83 $ ] followed by a point of a system for an arbitrary inertial observer @xmath84 .
any other inertial observer @xmath85 is related to the previous one by a transformation of a kinematical group such that their relative space - time measurements of any space - time event are given by @xmath86 where the functions @xmath87 and @xmath88 define the action of the kinematical group @xmath89 , of parameters @xmath90 , on space - time .
then the description of the trajectory of that point for observer @xmath85 is obtained from @xmath91.\ ] ] if we eliminate @xmath92 as a function of @xmath93 from the first equation and substitute into the second we shall get @xmath94 since observer @xmath85 is arbitrary , equation ( [ eq : rdet ] ) represents the complete set of trajectories of the point for all inertial observers .
elimination of the @xmath95 group parameters among the function @xmath96 and their time derivatives will give us the differential equation satisfied by the trajectory of the point .
this differential equation is invariant by construction because it is independent of the group parameters and therefore independent of the inertial observer . if @xmath89 is either the galilei or poincar group it is a ten - parameter group so that we have to work out in general up to the fourth derivative to obtain sufficient equations to eliminate the group parameters
. therefore the order of the differential equation is dictated by the number of parameters and the structure of the kinematical group .
let us consider first the case of the spinless point particle . in the non - relativistic case the relationship between inertial observers @xmath84 and @xmath85
is given by the action of the galilei group : @xmath97 in the relativistic case we have that the poincar group action is given by @xmath98 for the free spinless point particle it is possible to find a particular observer , the center of mass observer @xmath99 , such that the trajectory of the particle for this observer reduces to @xmath100.\ ] ] and therefore its trajectory for any other non - relativistic observer @xmath84 can be obtained from @xmath101 the trajectory of the point particle for the relativistic observer @xmath84 will be obtained from @xmath102 elimination of @xmath103 in terms of @xmath92 from the first equation of both ( [ eq : non ] ) and ( [ eq : rel ] ) and substitution into the second yields the trajectory of the point for an arbitrary observer , which in the relativistic and non - relativistic formalism reduces to @xmath104 elimination of group parameters @xmath27 , @xmath105 and @xmath106 by taking succesive derivatives yields the galilei and poincar invariant dynamical equation of a free spinless point particle @xmath107
we take spatial units such that the radius @xmath108 , and time units such that @xmath109 and therefore the velocity @xmath110 . for the center of mass observer ,
the trajectory of the charge of the electron is contained in the @xmath111 plane and it is expressed in 3-vector form as @xmath112 for the center of mass observer @xmath99 we get that @xmath113 for any arbitrary inertial observer @xmath84 we get @xmath114 we shall represent the different time derivatives by @xmath115 in this non - relativistic case @xmath116 , then , after using ( [ eq : rcm ] ) in some expressions we get the following derivatives @xmath117 therefore , the differential equation satisfied by the position of the charge of a non - relativistic electron and for any arbitrary inertial observer is @xmath118 we see that the motion is a helix because there is no @xmath119 term .
the center of mass position of the electron is defined as @xmath120 because it reduces to @xmath121 and @xmath122 for the center of mass observer , so that dynamical equations can be rewritten in terms of the position of the charge and the center of mass as @xmath123 our fourth - order dynamical equation ( [ eq : r4non ] ) can be split into two second order dynamical equations : a free equation for the center of mass and a central harmonic motion of the charge position @xmath3 around the center of mass @xmath124 of angular frequency 1 in these natural units .
the free dynamical equation @xmath125 is equivalent to @xmath126 , where @xmath127 is the linear momentum of the system .
then our free equations should be replaced in the case of an interaction with an external electromagnetic field by @xmath128,\quad { { \mbox{\boldmath$r$}}}^{(2)}={{\mbox{\boldmath$q$}}}-{{\mbox{\boldmath$r$ } } } , \label{eq : galilei}\ ] ] where in the lorentz force the fields are defined at point @xmath3 and it is the velocity of the charge that gives rise to the magnetic force term , while the second equation is left unchanged since it corresponds to the center of mass definition .
these equations are also obtained in the lagrangian approach @xcite by assuming a minimal coupling interaction and where we get @xmath129 , \label{eq : lag}\ ] ] which reduce to ( [ eq : galilei ] ) after the center of mass definition ( [ eq : cm ] ) . in order to determine the evolution of the system ,
initial conditions @xmath130 , @xmath131 , @xmath132 and @xmath133 , i.e. , the position of point @xmath3 and its derivatives up to order 3 evaluated at time @xmath134 must be given .
alternatively , if we consider that our fourth order differential equation ( [ eq : lag ] ) as the set of two second order differential equations ( [ eq : galilei ] ) , then we have to fix as initial conditions @xmath130 and @xmath131 as before and @xmath135 and @xmath136 , compatible with ( [ eq : cm ] ) , i.e. , the position and velocity of both points the center of mass and center of charge .
the advantage of this method is that we shall be able to analyse the evolution of a 2-electron system in section [ sec:2elec ] in terms of the center of mass initial position and velocity .
let us assume the same electron model in the relativistic case . since the charge is moving at the speed of light for the center of mass observer @xmath99 it is moving at this speed for every other inertial observer @xmath84 .
now , the relationship of space - time measurements between the center of mass observer and any arbitrary inertial observer is given by : @xmath137 with the shorthand notation for the following expressions : @xmath138 @xmath139 we obtain @xmath140 @xmath141 @xmath142 @xmath143 @xmath144 @xmath145 from this we get @xmath146 from equations ( [ eq : inv2])-([eq : inv4 ] ) we can express the magnitudes @xmath147 , @xmath148 and @xmath149 in terms of these scalar products between the different time derivatives @xmath150 .
the constraint that the velocity is 1 implies that all these and further scalar products for higher derivatives can be expressed in terms of only three of them .
if the three equations ( [ eq : r1])-([eq : r3 ] ) are solved in terms of the unknowns @xmath27 , @xmath151 and @xmath45 and substituded into ( [ eq : r4 ] ) , we obtain the differential equation satisfied by the charge position @xmath152 @xmath153 it is a fourth order ordinary differential equation which contains as solutions motions at the speed of light .
in fact , if @xmath154 , then by derivation we have @xmath155 and the next derivative leads to @xmath156 . if we take this into account and make the scalar product of ( [ eq : elecbuena ] ) with @xmath19 , we get @xmath157 , which is another relationship between the derivatives as a consequence of @xmath158
. it corresponds to a helical motion since the term in the first derivative @xmath19 is lacking .
the center of mass position is defined by @xmath159 we can check that both @xmath124 and @xmath160 vanish for the center of mass observer . then , the fourth order dynamical equation for the position of the charge can also be rewritten here as a system of two second order differential equations for the positions @xmath124 and @xmath3 @xmath161 a free motion for the center of mass and a kind of central motion for the charge around the center of mass . for the non - relativistic electron we get in the low velocity case @xmath162 and @xmath163 ,
the equations of the galilei case @xmath164 a free motion for the center of mass and a harmonic motion around @xmath124 for the position of the charge .
the free equation for the center of mass motion @xmath125 , represents the conservation of the linear momentum @xmath126 .
but the linear momentum is written in terms of the center of mass velocity as @xmath165 , so that the free dynamical equations ( [ eq : q2r2 ] ) in the presence of an external field should be replaced by @xmath166 where @xmath167 is the external force and the second equation is left unchanged because we consider , even with interaction , the same definition of the center of mass position .
@xmath168 we get @xmath169 and by leaving the highest derivative @xmath170 on the left hand side we finally get the differential equations which describe the evolution of a relativistic spinning electron in the presence of an external electromagnetic field : @xmath171\cdot{{\mbox{\boldmath$q$}}}^{(1)}\right)\right],\\ { { \mbox{\boldmath$r$}}}^{(2)}&=&\frac{1-{{\mbox{\boldmath$q$}}}^{(1)}\cdot{{\mbox{\boldmath$r$}}}^{(1)}}{({{\mbox{\boldmath$q$}}}-{{\mbox{\boldmath$r$}}})^2}\left({{\mbox{\boldmath$q$}}}-{{\mbox{\boldmath$r$}}}\right )
. \end{aligned}\ ] ]
if we have the relativistic and nonrelativistic differential equations satisfied by the charge of the spinning electrons we can analyse as an example , the interaction among them by assuming a coulomb interaction between their charges . in this way we have a system of differential equations of the form ( [ eq : galilei ] ) for each particle .
for instance , the external field acting on charge @xmath172 is replaced by the coulomb field created by the other charge @xmath173 at the position of @xmath172 , and simmilarly for the other particle .
the integration is performed numerically by means of the numerical integration program _
dynamics solver _ @xcite . in figure [ fig2 ]
we represent the scattering of two spinning electrons analysed in their center of mass frame .
we send the particles with their spins parallel and with a nonvanishing impact parameter .
in addition to the curly motion of their charges we can also depict the trajectories of their center of masses . if we compare this motion with the coulomb interaction of two spinless electrons coming from the same initial position and with the same velocity as the center of mass of the spinning electrons we obtain the solid trajectory marked with an arrow .
basically this corresponds to the trajectory of the center of mass of each spinning particle provided the two particles do not approach each other below compton s wave length .
this can be understood because the average position of the center of charge of each particle aproximately coincides with its center of mass and as far as they do not approach each other too much the average coulomb force is the same .
the difference comes out when we consider a very deep interaction or very close initial positions . in figure [ fig3 ]
we represent the initial positions for a pair of particles with the spins parallel .
the initial separation @xmath174 of their center of masses is a distance below compton s wave length .
we also consider that initially the center of mass of each particle is moving with a velocity @xmath175 as depicted .
that spins are parallel is reflected by the fact that the internal motions of the charges , represented by the oriented circles that surround the corresponding center of mass , have the same orientation .
it must be remarked that the charge motion around its center of mass can be characterised by a phase .
the phases of each particle are chosen opposite to each other .
we also represent the repulsive coulomb force @xmath176 computed in terms of the separation of the charges .
this interacting force @xmath176 has also been attached to the corresponding center of mass , so that the net force acting on point @xmath177 is directed towards point @xmath178 , and conversely .
we thus see there that a repulsive force between the charges represents an atractive force between their center of masses when located at such a short distance .
is computed in terms of the separation distance between the charges . ] in figure [ fig4 ] we depict the evolution of the charges and masses of this two - electron system for @xmath179 and @xmath180 during a short time interval .
figure [ fig5 ] represents only the motions of the center of masses of both particles for a longer time .
it shows that the center of mass of each particle remains in a bound region . the evolution of the charges is not shown in this figure because it blurs the picture but it can be inferred from the previous figure .
we have found bound motions at least for the range @xmath181 and velocity @xmath182 .
we can also obtain similar bound motions if the initial velocity @xmath175 has a component along the @xmath183 axis .
the bound motion is also obtained for different initial charge positions as the ones depicted in figure [ fig3 ] .
this range for the relative phase depends on @xmath174 and @xmath175 but in general the bound motion is more likely if the initial phases of the charges are opposite to each other .
we thus see that if the separation between the center of mass and center of charge of a particle ( _ zitterbewegung _ ) is responsible for its spin structure as has been shown in the formalism developped by the author then this atractive effect and also a spin polarised tunneling effect can be easily interpreted @xcite .
a bound motion for classical spinless electrons is not possible .
we can conclude that one of the salient features of this example is the existence from the classical viewpoint of bound states for spinning electron - electron interaction .
it is the spin structure which contributes to the prediction of new physical phenomena .
if two electrons have their center of masses separated by a distance greater than compton s wave length they always repel each other as in the spinless case .
but if two electrons have their center of masses separated by a distance less than compton s wave length they can form from the classical viewpoint bound states provided some initial conditions on their relative initial spin orientation , position of the charges and center of mass velocity are fulfilled .
the example analysed gives just a classical prediction , not a quantum one , associated to a model that satisfies dirac s equation when quantised .
the possible quantum mechanical bound states if they exist , must be obtained from the corresponding analysis of two interacting quantum dirac particles , bearing in mind that the classical bound states are not forbiden from the classical viewpoint . bound states for
a hydrogen atom can exist from the classical viewpoint for any negative energy and arbitrary angular momentum .
it is the quantum analysis of the atom that gives the correct answer to the allowed bound states .
the quantum mechanical analysis of a two electron system is left to a subsequent paper .
99 wigner e p 1939 _ ann . math .
( ny ) _ * 40 * 149 rivas m 1985 _ j. phys . a _ * 18 * 1971 1989 _ j. math .
phys . _ * 30 * 318 1994 _ j. math
. phys . _ * 35 * 3380 2001 _ kinematical theory of spinning particles _ , ( kluwer , dordrecht ) aguirregabiria j m 2002 _ dynamics solver _ , computer program for solving different kinds of dynamical systems , which is available from his author through the web site at the server of the theoretical physics dept .
of the university of the basque country , bilbao , spain . | we obtain by invariance arguments the relativistic and non - relativistic invariant dynamical equations of a classical model of a spinning electron .
we apply the formalism to a particular classical model which satisfies dirac s equation when quantised .
it is shown that the dynamics can be described in terms of the evolution of the point charge which satisfies a fourth order differential equation or , alternatively , as a system of second order differential equations by describing the evolution of both the center of mass and center of charge of the particle . as an application of the found dynamical equations ,
the coulomb interaction between two spinning electrons is considered .
we find from the classical viewpoint that these spinning electrons can form bound states under suitable initial conditions .
since the classical coulomb interaction of two spinless point electrons does not allow for the existence of bound states , it is the spin structure that gives rise to new physical phenomena not described in the spinless case .
perhaps the paper may be interesting from the mathematical point of view but not from the point of view of physics . |
the microscopic calculation of ground state energy and particle density of medium and heavy nuclei based on realistic nucleon - nucleon interaction requires the solution of a formidable many - body problem . for this reason
effective nucleon - nucleon interactions have been introduced , like the skyrme @xcite and gogny @xcite forces .
they simplify enormously the problem since , by construction , they must be used at the mean field level , and the calculation of the mean single particle potential and of the ground state energy becomes easily manageable .
the number of parameters which enter these effective forces is typically around ten and they are adjusted to reproduce finite nuclei and some equilibrium nuclear matter properties .
however , recently also data from a theoretically determined neutron matter equation of state ( eos ) @xcite have been used as input ( see also an older attempt in this direction in @xcite ) .
generally these forces give rise to an effective nucleon mass @xmath0 with typically @xmath1 in the non - relativistic framework . with these ingredients nuclear mean
field theories are very successful to describe nuclear properties as , e.g. binding energies , radii .
their use can be extended to the evaluation of more realistic single level schemes , of the nuclear excitation spectra , in particular giant resonances , of the fission barriers , of the nucleon - nucleus optical potential , and so on .
the price to be payed is that the connection with the bare nucleon - nucleon interaction is not apparent .
the first attempt to connect the effective nn interaction and the underlying bare interaction was the density matrix expansion of negele and vautherin @xcite , based on the assumption that the effective nn interaction can be identified essentially with the brueckner g - matrix calculated in nuclear matter and on the gradient expansion of the non - local one - body density matrix .
see also another attempts in the same direction in ref .
this approach was recently generalized to include also the vector ( spin ) part of the density matrix at the same level of accuracy as the scalar part @xcite .
the challenging program of deriving the effective interaction to be used in nuclear structure studies from the bare nn interaction has still to be completed .
an approach similar to the skyrme - like one is the energy density functional ( edf ) approach , where the basic quantity to start with is directly the functional that expresses the energy in terms of the matter density and its gradients .
the functional must be minimized to obtain the actual ground state energy and matter density profile . following the method of kohn and sham @xcite , developed in atomic , molecular and solid state physics , the minimization procedure can be performed by introducing a set of auxiliary single particle states and taking for the kinetic energy the slater form in this basis . in this way
the minimization procedure gives hartree - like equations for the single particle states , where the interaction part includes in an effective way the overall exchange and correlation contributions .
the latter , in condensed matter physics , is taken from accurate calculations for the homogeneous electron system or in a purely phenomenological fashion .
this is not the usual way of proceeding in nuclear physics , although it exists , to our knowledge , an earlier attempt in this direction , see ref .
@xcite . in this paper we will discuss to
what extent the edf approach in nuclear structure can be based on microscopic many - body results on nuclear matter equation of state ( eos ) and which accuracy can be reached . to this purpose
we will rely on recent achievements along this line of research and we will analyze the main open problems that hinder the development of the microscopic many - body theory of the nuclear edf .
despite the fact that the applicability of the ks - dft approach to self - bound systems , as nuclei , is not obvious @xcite , it is commonly believed that the basis of ks - dft lies in the hohenberg - kohn ( hk ) theorem @xcite , which states that for a fermi system , with a non - degenerate ground state , the total energy can be expressed as a functional of the density @xmath2 only .
such a functional reaches its variational minimum when evaluated with the exact ground state density .
furthermore , in the standard ks - dft method one introduces an auxiliary set of @xmath3 orthonormal single particle wave functions @xmath4 , where @xmath3 is the number of particles , and the density is assumed to be given by @xmath5 where @xmath6 and @xmath7 stand for spin and iso - spin indices .
the variational procedure to minimize the functional is performed in terms of the orbitals instead of the density . as in condensed matter and atomic physics the hk functional @xmath8
is split into two parts : @xmath9+w[\rho]$ ] @xcite .
the first piece @xmath10 corresponds to the uncorrelated part of the kinetic energy and within the ks method it is written as @xmath11 the other piece @xmath12 $ ] contains the potential energy as well as the correlated part of the kinetic energy .
then , upon variation , one gets a closed set of @xmath3 hartree - like equations with an effective potential , the functional derivative of @xmath12 $ ] with respect to the local density @xmath2 .
since the latter depends on the density , and therefore on the @xmath13 s , a self - consistent procedure is necessary .
the equations are exact but they only can be of some use if a reliable approximation is found for the otherwise unknown density functional @xmath12 $ ] .
it has to be stressed that in the ks - dft formalism the exact ground state wave function is actually not known , the density being the basic quantity . in nuclear physics , contrary to the situation in condensed matter and atomic physics , the contribution of the spin - orbit interaction to the energy functional is very important .
non - local contributions have been included in dft in several ways already long ago ( see @xcite for a recent review of this topic ) .
consequently , the spin - orbit part also can be split in an uncorrelated part @xmath14 plus a remainder .
the form of the uncorrelated spin - orbit part is taken exactly as in the skyrme @xcite or gogny forces @xcite .
we thus write for the functional in the nuclear case @xmath15 , where we explicitly split off the coulomb energy @xmath16 because it is a quite distinct part in the hamiltonian .
it shall be treated , as usual , at lowest order , i.e. the direct term plus the exchange contribution in the slater approximation , that is @xmath17 , and @xmath18 with @xmath19 and @xmath20 the proton / neutron density .
let us now discuss the nuclear energy functional contribution @xmath21 $ ] which contains the nuclear potential energy as well as additional correlations .
we shall split it in a finite range term @xmath22 $ ] to account for correct surface properties and a bulk correlation part @xmath23 $ ] that we take from a microscopic infinite nuclear matter calculation @xcite .
thus our final ks - dft -like functional reads : @xmath24 for the finite range term we make the simplest phenomenological ansatz possible @xmath25 & = & \frac{1}{2}\sum_{t , t'}\int \int d^3r d^3r'\rho_{t}({\bf r } ) v_{t , t'}({\bf r}-{\bf r'})\rho_{t'}({\bf r ' } ) \nonumber \\ & - & \frac{1}{2}\sum_{t , t ' } \gamma_{t , t'}\int d^3r { \rho_{t}({\bf r } ) } \rho_{t'}({\bf r } ) \label{eq : eq7}\end{aligned}\ ] ] with @xmath26 proton / neutron and @xmath27 the volume integral of @xmath28 .
the substraction in ( [ eq : eq7 ] ) is made in order not to contaminate the bulk part , determined from the microscopic infinite matter calculation .
finite range terms have already been used earlier , generalizing usual skyrme functionals ( see e.g @xcite ) . in this study , for the finite range form factor @xmath28 we make a simple gaussian ansatz : @xmath29 .
we choose a minimum of three open parameters : @xmath30 , and @xmath31 .
the only undetermined and most important piece in ( [ eq : eq6 ] ) is then the bulk contribution @xmath32 .
as already mentioned , we obtain @xmath32 from microscopic infinite matter calculations , using a realistic bare force , together with a converged hole - line expansion @xcite .
we first reproduce by interpolating functions the _ correlation _ part of the ground state energy per particle of symmetric and pure neutron matters , and then make a quadratic interpolation for asymmetric matter
. finally the total correlation contribution to the energy functional in local density approximation reads : @xmath33 = \int d^3r \big [ p_s(\rho ) ( 1 - \beta^2 ) + p_n ( \rho)\beta^2 \big ] \rho \label{eq : eq4a}\ ] ] where @xmath34 and @xmath35 are two interpolating polynomials for symmetric and pure neutron matter , respectively , at the density @xmath36 , and @xmath37 is the asymmetry parameter .
the interpolating polynomial for symmetric matter has been constrained to allow a minimum exactly at the energy @xmath38 = - 16.00 mev and fermi momentum @xmath39 = 1.36 @xmath40 , i.e. @xmath41= 0.16 @xmath42 .
this is within the uncertainty of the numerical microscopic calculations of the eos .
it has to be stressed that the use of a polynomial in density is just for practical reasons .
the constrained fit was performed by keeping the eos as smooth as possible , thus allowing for some very small deviations from the microscopic calculations below saturation density .
an interpolating fit which goes exactly through the calculated eos , as performed in @xcite , gives a not good enough saturation point ( typically e / a = -15.6 mev , @xmath39=1.38 @xmath40 ) .
as discussed in @xcite , the low density behavior of the nuclear matter eos is quite intricate and usually not reproduced by skyrme and gogny functionals ( see also ref .
@xcite ) , missing quite a substantial part of binding .
we show our eos for nuclear and neutron matter in fig .
since we want to construct the edf , as much as possible , on the basis of the microscopic calculations , the bulk part @xmath32 of the functional , directly related to bare nn and tbf ( three - body forces ) , is determined once and for all and we will use it in ( 3 ) together with lda .
[ fig : eos ] the only open parameters are , therefore , the ones contained in the finite range surface part , eq.([eq : eq7 ] ) , and the strength of the spin - orbit contribution , that we fit to reproduce finite nuclei properties .
we , thus , follow _ exactly _ the strategy employed in condensed matter . on the contrary , in nuclear physics , almost exclusively a different strategy is usually adopted ( see , however , ref.@xcite with some ingredients similar as in the present approach ): functionals like the one of ( [ eq : eq6 ] ) , i.e. bulk , surface , etc .
, were globally parametrized with typically of the order of ten parameters which , then , were determined fitting _ simultaneously _ some equilibrium nuclear matter ( binding energy per particle , saturation density , incompressibiliy , etc . ) and finite nuclei properties . however , in this way , bulk and surface are not properly separated and early attempts used to miss important infinite nuclear matter properties , as , e.g. stability of neutron matter at high density @xcite and other stability criteria .
modern skyrme forces , like the saclay - lyon ( sly ) ones , explicitly use the high density part ( @xmath43 ) of microscopic neutron matter calculations for the eos in the fitting procedure @xcite and thus avoid collapse .
therefore , modern functionals usually reproduce reasonably well microscopically determined eos for neutron and nuclear matter @xcite .
examples are , among others , the sly - forces @xcite ( see fig.1 in @xcite ) and the fayans functional df3 ( see e.g. fig . 3 in @xcite ) .
in this presentation we follow the just discussed alternative approach , that is different on a qualitative level from the usual and allows , via the fit , to reproduce very accurately the microscopic infinite matter results in the whole range of densities considered . this may be important for surface properties and neutron skins in exotic nuclei , what shall be investigated in the future . for open shell nuclei
, we still have to add pairing .
the formal generalization of the rigorous hk theorem to paired system has been given in ref .
@xcite . in the present work
our main objective is to discuss the ks - dft scheme for the non pairing part , thus we add pairing in a very simple way within the bcs approach . for this
we simply take the density dependent delta force defined in ref @xcite for @xmath44 with the same parameters and in particular with the same cutoff .
as far as this amounts to a cutoff of @xmath45 10 mev into the continuum for finite nuclei , we have to deal with single - particle energy levels lying in the continuum .
we have simulated it by taking in the pairing window all the quasi - bound levels , i.e. the levels retained by the centrifugal ( neutrons ) and centrifugal plus coulomb ( protons ) barriers .
this treatment of the continuum works properly , at least for nuclei not far from the stability valley as it has been extensively shown in @xcite .
in this way we obtain two - neutron ( @xmath46 ) and two - proton ( @xmath47 ) energy separations for magic proton and neutron numbers in quite good agreement with the experiment ( see also below ) . in our calculations
the two - body center of mass correction has been included in the self - consistent calculation using the pocket formula , based on the harmonic oscillator , derived in ref.@xcite which nicely reproduces the exact correction as it has been shown in @xcite .
our functional is now fully defined and , henceforth , we call it bcp - functional . preliminary results based on this functional have been presented in ref .
in this section we discuss the fitting protocol to determine the open parameters of our functional and the accuracy that can be reached following the scheme described in the previous section . to this purpose
we make a comparison between our recent results and the ones that can be obtained with the gogny functional , which is one of the most accurate phenomenological functional . in this paper
we follow a novel strategy , recently proposed @xcite , for the fitting procedure to obtain the set of parameters for the surface and spin - orbit part of the functional , that we call the bcp09 functional .
contrary to our previous works we no longer use the binding energies of spherical nuclei for the fit but , following the suggestion of ref .
@xcite , we use a set of deformed nuclei carefully chosen in the rare earth , actinide and super - heavy regions instead .
the underlying philosophy behind this choice @xcite is that for spherical nuclei the fluctuations in the two most relevant collective variables , namely , the quadrupole deformation and the pairing correlations are large ( i.e. the minimum in those variables is broad in some sense ) and therefore correlations beyond mean field are important and difficult to evaluate . on the other hand , for deformed nuclei
the minimum as a function of the quadrupole degree of freedom is stiffer than its counterpart in spherical nuclei and the additional correlations ( rotational energy correction mainly ) are not so difficult to compute in a mean field model . in order to fit the three free parameters of the surface part ( @xmath48 , @xmath49 and @xmath31 ) of bcp and the spin - orbit strength @xmath50 we have taken as in @xcite 84 well deformed nuclei in the rare earth , actinide and super - heavy regions where the experimental binding energies are known .
the four parameters are determined by minimizing the mean square root deviation of the binding energy @xmath51 .
the theoretical binding energy has been computed as the hartree- fock- bogoliubov ( hfb ) mean field energy plus a rotational energy correction , which differs from the standard one @xcite by a phenomenological factor to account for the approximate calculation of the yoccoz moment of inertia .
we also consider an additional correction factor ( not relevant in the present calculation ) to deal with the weak deformation regime ( see , for instance , @xcite )
. also a correction to the binding energy ensuing from the finite size of the harmonic oscillator basis used , as estimated in @xcite , has been included .
axial symmetry has been preserved in the hfb calculation as the nuclei chosen are not expected to develop triaxiality . on the other hand reflection symmetry
is allowed to be broken in the solution of the hfb equation , to allow for octupole deformation in the ground state which is relevant for a few of the actinides considered .
the calculation of the 84 ground states for a set of parameters can be carried out in a powerful personal computer in around half an hour . as a consequence of this figure ,
an unconstrained and blind search of the four parameters leading to the absolute minimum of @xmath52 is a task out of the scope of the present exploratory considerations .
fortunately , the expression for the binding energy depends linearly on three out of the four free parameters ( @xmath48 , @xmath49 and @xmath50 ) and the method suggested by bertsch et al .
@xcite in those situation should work well .
we have implemented bertsch s procedure for the three parameters @xmath48 , @xmath49 and @xmath50 and performed systematic calculations for as a function of the other parameter @xmath31 .
usually , for a given @xmath31 value and with a reasonable choice of starting parameters , the linearized bertsch s method leads to a local minimum of @xmath51 in a couple of iterations .
the reason is that the correlation matrix of bertsch s method has in our case two eigenvalues which are two to three orders of magnitude smaller than the remaining one .
this fact points to the existence of only one free parameter ( a linear combination of @xmath48 , @xmath49 and @xmath50 with weights corresponding to the components of the eigenvector of the largest eigenvalue ) . at this point
it has to be mentioned that the fact that the expression of the binding energy is linear in @xmath48 , @xmath49 and @xmath50 by no means imply the value of binding energy to be globally a linear function of those quantities .
the reason is that the binding energy is obtained after a self - consistent hfb calculation and the self - consistency breaks the linear dependence of the wave function on the parameters mentioned above . as a consequence ,
bertsch method is only valid locally what implies that the procedure will end up in different minimum depending upon the values of the initial parameters .
a careful analysis of the results obtained so far indicates that the spin - orbit strength is the most sensitive starting parameter and therefore a search on this degree of freedom should be carried out in addition to the search on @xmath31 . using this new protocol
we have performed a fit using for the bulk density dependent part of the functional new interpolating polynomials for the nuclear matter equation of state in a wider range of density with respect to the previous ones ( bcp1 and bcp2 in refs .
the new fit has still a binding energy per nucleon @xmath53 mev at saturation density @xmath54 fm @xmath55 ( for a compressibility of 220 mev ) . for the spin - orbit strength we choose to stay around the @xmath56 mev value and perform a minimization of @xmath51 for different values of @xmath31 from 0.85 fm up to 1.00 fm .
the minimum value of @xmath51 was obtained for @xmath57 fm .
the values of the other parameters are @xmath58 mev , @xmath59 mev and @xmath60 mev @xmath61 for a value of @xmath51 of 0.545 mev . in order to check for the suitability of the spin - orbit strength value used
we performed another two minimizations of @xmath51 fixing @xmath31 at 0.9 fm but taking @xmath62 mev @xmath61 and 102.51 mev @xmath61 .
the values of @xmath51 obtained are 0.569 mev and 0.566 mev respectively , supporting the assignment of @xmath63 mev @xmath61 as the value leading to the lowest @xmath51 ( this check is incomplete as a more extensive search in the two parameters should be performed ; work in this direction is in progress and will be reported in a near future ) .
16 mev , @xmath57 fm that yield @xmath64 mev are plotted as full ( blue ) dots , whereas results for @xmath6516 mev , @xmath66 fm that yield @xmath67 mev are plotted as open ( green ) circles .
the group of points to the left corresponds to the rare earth isotopes whereas the one to the right are for the actinides and superheavy.[fig : sigmabcp09 ] ] in fig .
[ fig : sigmabcp09 ] we show the individual differences in binding energy ( @xmath68 ) between the theoretical results and the experimental ones for the 84 nuclei considered .
the quantity @xmath69 has been plotted as a function of the number of neutrons @xmath70 and the values corresponding to the same isotope are joined with lines .
two sets of parameters have been considered , namely one with @xmath57 fm producing the lowest value of @xmath51 ( full dots ) and other with a slightly higher value @xmath67 mev and obtained fixing @xmath66 fm ( circles ) .
it is worth pointing out that the absolute value of @xmath71 never exceeds 1.2 mev .
we also observe that the change in parameters only produce minor changes in the @xmath71 s .
finally , we notice that the actinides and superheavies look a little bit too underbound as compared to the rare earth isotopes .
finding the origin of such a relative underbinding will help to improve the quality of the fit reducing the value of @xmath51 .
although the value of @xmath51 was obtained by considering an optimal set of deformed nuclei ( those with a well established deformation ) it is very likely that the @xmath51 value for a complete set of nuclei from proton drip line to neutron drip line will be promising . in order to compare the results obtained with bcp 09 with others of other interactions we have performed calculations with the gogny force and the most recent parametrization d1 m that was made up with the idea of producing a mass table .
unfortunately , the details of how the theoretical binding energy was obtained are not detailed enough as to allow a reproduction of them ( the main uncertainty is in the quadrupole motion zero point energy correction ) .
we have decided to use the same protocol as with bcp 09 , that is , the rotational energy correction and the finite size of the basis effect are included , and we have left out the zero point energy correction mentioned above . as a consequence ,
our values for @xmath51 of d1 m are too high and in order to make a more fair comparison we just shifted all the binding energies computed by a constant quantity ( 2.8 mev ) obtained as to minimize @xmath51 . with this readjustment
we obtain for d1 m and the 84 nuclei considered the value @xmath72 mev which is slightly better than our value but , on the other hand , of similar quality to the one of the bcp 09 functional . in fig .
[ fig : sigmabcp09d1 m ] we have plotted the individual values of @xmath71 both for bcp 09 and gogny d1 m .
we notice how the behavior of @xmath71 for the gogny d1 m as a function of @xmath70 for each isotopic chain is different from the one of bcp 09 ; decreasing with increasing @xmath70 in the case of gogny d1 m whereas it is increasing with increasing @xmath70 for bcp 09 .
it is also worth to point out that gogny d1 m force produces results in the actinide and super - heavy nuclei which are more spread out than the ones of the bcp 09 functional . by comparing both results
we can conclude that , at least for the 84 deformed nuclei considered , the performance of bcp 09 is as good as the one of gogny d1 m and at a fraction of the computational cost ( the absence of exchange terms and non local pairing in bcp makes the numerical evaluation of the quantities entering the hfb equation much faster than for the gogny d1 m case ) .
this is an encouraging starting point for a more detailed study of the viability of bcp 09 in the description of low energy nuclear structure .
but this time the bcp results with @xmath6516 mev , @xmath57 fm that yield @xmath64 mev are compared to those of the gogny d1 m force . see text for further details .
[ fig : sigmabcp09d1 m ] ] next we want to explore the ability of our proposed bcp09 energy density functional to describe nuclear ground - state properties of other nuclei not included in the fitting protocol used to obtain the free parameters that describe the surface and spin - orbit contributions to the functional .
to this end we have computed the ground - state energies of 161 even - even and 306 odd spherical nuclei .
these nuclei are chosen to be spherical according to the deformation properties of the compilation of mller and nix @xcite . to deal with odd nuclei ,
we have used the blocking approach on top of the bcs calculation .
the differences @xmath73 , where @xmath74 are the theoretical predictions of the bcp09 functional and @xmath75 the experimental values taken from @xcite , are displayed in fig .
[ fig : diffbsph ] . the agreement found between the theoretical prediction and the experimental values is fairly good finding an energy rms @xmath76 1.3 mev .
it should be mention that both , even - even and odd nuclei , give basically the same rms when they are considered separately . from the figure
we can see that the theoretical ground - state energies are scattered , rather symmetrically , around @xmath77 within a window of about @xmath78 3 mev with the only exception of the nucleus @xmath79er . combining these values for 467 spherical nuclei with the results obtained for the 84 deformed nuclei used in the fitting protocol , we find a global rms @xmath801.2 mev , a value that can be improved if the analysis of the ground - state energies include more deformed nuclei . in fig .
[ fig : diffrsph ] we show the differences between the calculated and experimental charge radii @xmath81 . in this figure
we have displayed the differences @xmath82 for 88 even - even and 111 odd spherical nuclei for which the experimental charge radii are known @xcite . from this figure
we see that the bcp09 energy density functional predicts , on the average , smaller charge radii than the experimental values . with few exceptions , the theoretical charge radii
are scattered within a window of 0.04 fm around an average values @xmath82 = -0.02 fm . the global quality of this fit of the charge radii is given by rms @xmath83=0.030 fm and again the rms of the even - even and odd nuclei are basically the same .
this global rms can be compared with one provided by the hfb-8 model @xcite which is @xmath83=0.0275 fm , however computed using 782 experimental data . as compared with the results provided by the earlier bcp2 functional , that included some experimental charge radii in the fit , the @xmath83 values
are roughly the same , but in the case of the bcp2 functional the theoretical charge radii are rather overestimated , at least for light nuclei . after having defined a performing functional in the realm of nuclear masses one has to check its suitability in describing other nuclear properties like deformations .
a detailed study of deformations with bcp 09 is underway and will be reported in the future , but in order to give a taste of the results already obtained we show in fig .
[ fig : mfprop240pu ] the typical outcome of a fission barrier calculation for the nucleus @xmath84pu that has also been discussed in ref .
@xcite for early versions of the bcp functional . in fission studies the energy of the system obtained in the hfb framework
is obtained as a function of the quadrupole moment by constraining the hfb solution . in this way a potential energy curve ( pec ) is obtained that in the present case is displayed in the lower panel of fig .
[ fig : mfprop240pu ] for the bcp1 and bcp09 functionals and the gogny d1s force .
the bcp1 and bcp09 curves have been shifted by 0.65 mev and 4.5 mev respectively as to make the first minimum at @xmath85 b ( @xmath86 0.26 ) coincide in energy .
after the first ( ground state ) minimum there is a barrier ( first fission barrier ) that is higher for d1s than for the bcp functionals .
the height of this barrier is known to be very sensitive to triaxial effects which are still not accounted for in the present calculation .
after the first fission barrier there is an additional excited local minimum at @xmath87 b ( @xmath860.73 ) that corresponds to the fission isomer .
the excitation energy of the fission isomer obtained with the two bcp functionals is more or less the same with a value of 3.5 mev .
for the gogny d1s the result is slightly higher ( around 4.5 mev ) .
the following maximum corresponds to the second fission barrier which is definitely much higher for gogny d1s than for the bcp functional .
after the second fission barrier the typical decrease of the energy consequence of the coulomb repulsion between the nascent fragments ( the real shape of the nucleus is depicted at the corresponding @xmath88 value as a contour plot of the real density at a value of the contour of 0.08 @xmath42 ) . as the two fragments are still connected by the neck the nuclear part of the interaction still has an effect , explaining the differences observed between the three calculations .
finally , in the upper panel of fig .
[ fig : mfprop240pu ] the octupole ( @xmath89 ) and hexadecapole ( @xmath90 ) moments of the mass distribution obtained with the three functionals / interactions are shown .
the values lie on top of each others indicating ( as the multipole moments considered are the most relevant ) that the shapes of the nucleus along the whole fission path are essentially the same for the three calculations .
for the two bcp functionals and gogny force used in the calculations is shown . in the upper panel ,
the octupole and hexadecapole moments are displayed again for the three interactions / force used .
the lines lie on top of each other and are hardly distinguisable .
see text for further details .
[ fig : mfprop240pu ] ]
the problems to be clarified in the edf approach can be in general classified in two categories .
on one hand , the use of the nuclear matter eos must be taken with care since the microscopic calculations can not reach the accuracy usually required in nuclear structure calculations and contains intrinsic and unavoidable uncertainties discussed below . on the other hand ,
finite size effects must be included in the functional with a minimal set of parameters .
the procedure of doing that is not straightforward and requires extensive analysis . following the ks scheme
, the nuclear eos calculated microscopically should be included in the edf mainly by an analytic fit that reproduces accurately both the nuclear matter and neutron matter results . in this way the saturation point , as well the whole density dependence of the binding energy , can be incorporated in the edf .
however , the microscopic calculations can not be considered without any uncertainty .
besides the point of numerical accuracy , it is well known that it is necessary to introduce three - body forces in the microscopic calculations in order to reproduce the correct saturation point , which however contains necessarily a phenomenological uncertainty .
therefore the tbf can be adjusted to reproduce different saturation points .
furthermore the microscopic data do not fix uniquely the fitted eos .
we found that the quality of the nuclear data fitting is very sensitive to the saturation point of the interpolated eos .
the optimal value of the energy per particle at saturation seems to be close to -16 mev , but even a shift as small as 0.1 mev of this quantity can deteriorate appreciably the quality of the nuclear data fit .
this should be a general feature , independent of the particular edf that is used .
less sensitivity seems to be present for the saturation density . for the eos obtained from the bhf theory ,
the tbf contribution below saturation is quite small and the bhf is expected to be accurate at low density , so that the main problem can be considered as how to fix the saturation point .
of course the just discussed sensitivity on the saturation point could be considered a not serious problem , in view of the phenomenological uncertainty that affects its position .
all these problems appear absent for the neutron matter eos , for which the tbf contribution is smaller and correlation energy is substantially reduced .
a separate discussion is needed for the type and strength of the tbf that are used in microscopic nuclear matter calculations .
the numerical calculations of the binding energy of nuclear few - body systems , like triton and alpha particle , can be performed numerically very accurately .
if realistic two - body forces , like the argonne v@xmath91 potential , are used the binding energies turn out systematically underestimated . to remedy to this drawback
usually one adds a semi - phenomenological tbf , like the urbana model @xcite , that is adjusted to reproduce the experimental binding energies .
the unpleasant discovery is that the use of the same tbf in microscopic calculations for nuclear matter does not produce a good saturation point .
the tbf appear too repulsive . indeed ,
in general semi - phenomenological tbf s contain a repulsive and attractive part , and the standard procedure is to reduce the repulsive component in order to get a good saturation point . at phenomenological level
this means that one needs at least a four - body force in a non - relativistic many - body scheme .
it has to be mentioned in this respect that non - local two - body forces @xcite which are able to reproduce the binding energy of few - nucleons systems fail to reproduce the correct saturation point , that then turns out to be at too high density and too low energy @xcite .
recently @xcite it has been shown that one can get a reasonable saturation point by keeping the tbf obtained by fitting few - body systems if for the effective two - body force the so - called v@xmath92 interaction is used .
the latter is obtained by projecting out the high momenta of the bare nn interaction and renormalizing accordingly the interaction at low momenta .
the resulting interaction is then phase - equivalent to the original bare nn interaction up to the momentum cutoff , but it is much softer , and therefore it can be treated perturbatively .
nuclear matter calculations with v@xmath92 are not saturating , most probably because they misses the effects of the pauli principle and of the self - consistent single particle potential , which are distint features of the g - matrix .
the final saturation point is a consequence of the compensation between the too large attraction of v@xmath92 and the too large repulsion of the tbf .
all these uncertainties are of course embodied in the final eos and affect the detailed properties of the resulting edf .
unfortunately there is a more basic problem for the low density eos . at low enough density symmetric matter
is expected to be unstable towards cluster formation .
the picture of an homogeneous matter can not be kept at very low density .
the real eos is therefore altered by the appearance of clusters of different sizes .
however the eos which includes cluster formation can not be used in the edf , since such a proliferation of light nuclei can not be present in the low density region of finite nuclei .
this is a general problem for any edf .
the absence of clusters at the nuclear surface is due to the small value of the diffusion of the density profile , which prevents any long range correlation to dominate .
this fact offers a partial justification of taking the homogeneous matter eos at the bhf level .
indeed , short range correlations are well taken into account by the bhf procedure , and therefore the corresponding eos should be able to describe the properties of the bulk contribution to the nuclear edf even in the surface region .
this problem still needs further analysis before it can be considered satisfactorily solved . in the standard ks procedure
the kinetic energy contribution is taken at the independent particle level with an effective mass equal to the bare one .
this is in agreement with the bhf scheme , where the kinetic energy part is also kept at the free value and the whole correlation contribution is included in the interaction energy part coming from the g - matrix .
this means that also the correlated part of the kinetic energy is included in the g - matrix contribution .
indeed the momentum dependence of the g - matrix is clearly the origin of the deviation of the effective mass value from the bare one .
it would not be difficult to separate this effect from the g - matrix contribution , so that the kinetic energy would include a ( density dependent ) effective mass , while leaving the rest of correlations as a genuine interaction energy .
this leaves some freedom to the way the nuclear matter eos is actually included in the edf , which allows to go beyond the usual ks scheme .
it can be expected that different procedures are not necessarily equivalent .
this problem has still to be analyzed in detail .
the density matrix expansion of negele and vautherin @xcite suggests that the surface contribution to the binding energy of nuclei may also be extracted from nuclear matter calculations .
this would be very welcome , since this would further reduce the number of adjustable parameters and link the functional even more to the microscopic approach .
essentially only the spin - orbit and pairing contributions would then need some phenomenological adjustments .
however , the g - matrix also has spin - spin interaction terms and , in principle , the spin orbit also could be extracted if the vector part of the one - body density matrix is properly taken into account @xcite .
to what extent such a goal can be achieved remains an open problem .
let us , however , point to some possible improvements over the past procedures . in first place
we want to point out that it is our believe that neither the dme of negele- vautherin , nor the one of campi - bouissy @xcite are the optimal procedures .
we have long standing experience with the semiclassical dme of wigner and kirkwood which is based on a systematic expansion in powers of @xmath93 .
most accurate results are obtained , at least for the scalar part of the one- body density matrix , with this method @xcite .
a second problem stems from the way how the g - matrix of an infinite matter calculation is used for a finite nucleus .
the standard procedure is the lda , that is one replaces the @xmath39 dependence by a density dependence via the standard relation @xmath94 .
the @xmath39 dependence enters mostly into the pauli operator @xmath95 .
however , dependence also can be in the self consistent single particle energies .
we want to point out that just replacing @xmath39 by @xmath96 is not necessarily the best procedure .
our argument goes as follows .
the two particle propagator entering the g - matrix equation @xmath97 [ e - \hat h_1 - \hat h_2 -i\eta]^{-1}\ ] ] has an obvious semiclassical , i.e. @xmath98 limit , namely we just replace the single particle hamiltonians @xmath99 by their classical counterparts @xmath100 with @xmath101 the phase space variables of position and momenta . with this approximation
the two particles move according to their respective classical hamiltonians , each .
supposing for simplicity a local mean field @xmath102 ( the argument can be generalized to nonlocal mean field potentials ) , we see that in the semiclassical two body propagator two local chemical potentials appear : @xmath103 and @xmath104 ( @xmath105 ) and , therefore , two fermi momenta and , thus , two densities at positions @xmath106 appear in the semiclassical propagator and correspondingly in the g - matrix .
since the effective range of the interaction is over 2 fm wide , this non locality in the densities may be of quite some importance .
one may call this the non local density approximation ( nlda ) .
the standard lda is recovered in putting @xmath107 .
this nlda effect may be more important , or at least of equal importance , than to keep gradient expansions of the density itself
. it could be interesting to test this .
since finite size effects are surface effects , one may argue that only relatively low densities will be involved .
it has recently been shown , that at least in neutron matter for low densities a separable force works very well for the g - matrix @xcite , reducing very much the numerical effort .
so eventually one could combine the nlda approach with a separable force to get to a surface term .
of course , one will not imagine that with such a procedure one will hit the good result right on the spot .
but even if the result is only semi quantitatively correct , it still may lead to a reduction of the open parameters and to more insight into the underlying physics . this may be then a first step in the direction to get everything from the microscopic input , i.e. from the underlying bare forces , also for finite nuclei .
we have presented a formulation of the energy functional method that enables to keep , to a certain extent , the connection with the nuclear matter equation of state as calculated microscopically from many - body theory and realistic nucleon - nucleon interactions .
the functional is constructed following closely the kohn and sham method , where the bulk part of the functional is taken directly from the microscopic calculations and kept fixed in the fitting procedure of the adjustable parameters of the functional .
the latter include the surface part of the functional and the spin - orbit strength , ending up with a total of four free parameters .
the pairing interaction and the corresponding strength has been taken from the simplest standard choices . despite the analysis
was not carried out up to the optimal level , leaving space for refinements , the accuracy of the results can compete with the one obtained with the best purely phenomenological functional like gogny .
these promising results open the possibility of building an energy density functional closely connected with the bare nucleon - nucleon forces .
developments in this direction could be obtained along the density matrix expansion method of ref .
@xcite or its refinements , which could further reduce the number of adjustable parameters .
however , we pointed out the various open problems that must be solved before this project can be completed .
the incorporation of the nuclear eos in the density functional can be done in different ways , and this leaves some ambiguity in the direction of a universal nuclear functional .
furthermore the very low density part of the eos can not be taken literarily from microscopic calculations of homogeneous nuclear matter .
in fact cluster formation is expected to dominate this region of the eos .
this appears a serious problem for any density functional theory which is constructed in such a way to be compatible with the nuclear matter case .
further studies and analysis are needed to proceed further in this project on the nuclear density functional .
work supported in part by micinn ( spain ) grants fpa2007 - 66069 and fpa2008 - 03865-e / in2p3 , and by the consolider - ingenio 2010 program ( spain ) cpan , csd2007 - 00042 .
x. v. also acknowledges the support from fis2008 - 01661 ( spain and feder ) and 2009sgr-1289 ( spain ) .
this work was supported by compstar , a research networking programme of the european science foundation . | in recent years impressive progress has been made in the development of highly accurate energy density functionals , which allow to treat medium - heavy nuclei . in this approach one
tries to describe not only the ground state but also the first relevant excited states . in general
, higher accuracy requires a larger set of parameters , which must be carefully chosen to avoid redundancy . following this line of development , it is unavoidable that the connection of the functional with the bare nucleon - nucleon interaction becomes more and more elusive . in principle
, the construction of a density functional from a density matrix expansion based on the effective nucleon - nucleon interaction is possible , and indeed the approach has been followed by few authors .
however , to what extent a density functional based on such a microscopic approach can reach the accuracy of the fully phenomenological ones remains an open question .
a related question is to establish which part of a functional can be actually derived by a microscopic approach and which part , on the contrary , must be left as purely phenomenological . in this paper
we discuss the main problems that are encountered when the microscopic approach is followed . to this purpose
we will use the method we have recently introduced to illustrate the different aspects of these problems .
in particular we will discuss the possible connection of the density functional with the nuclear matter equation of state and the distinct features of finite size effects proper of nuclei . |
time - resolved pump / probe experiments have seen a rebirth in recent years as timescales have been pushed well into the femtosecond ( and sometimes attosecond ) range , and as many different types of experiments can now be performed in an ultrafast pump / probe format .
much of the experimental work , ranging from ultrafast optical studies to angle - resolved photoemission ( arpes ) have been centered on examining the behavior of charge density wave materials @xcite . in this paper
, we will review work on theoretical descriptions of pump / probe time - resolved photoemission spectroscopy on charge - density - wave ( cdw ) materials .
because this is a short review , we will focus entirely on our work , which describes many of the experimental features seen in recent studies .
pump / probe photoemission experiments really came to life when it was recognized that one could period quadruple a 1.5 ev light pulse to create 6 ev pulses that also are ultrashort ( typical widths are in the few 10s of femtoseconds ) .
the higher - energy pulses are sufficient to photoemit electrons from part of the brillouin zone ( bz ) in materials with small workfunctions . the experimental apparatus for this
was then applied to a wealth of different materials , with a large focus on cdw systems like tas@xmath0 , tise@xmath0 and tbte@xmath1 .
these materials have complex ground states , and often have a number of different competing phases that can be accessed by changing temperature or pressure .
tas@xmath0 was the first material studied @xcite .
in the 1 t phase , this material goes through a range of different phase transitions , starting with an incommensurate cdw at high temperature , passing to a nearly commensurate cdw , and then a commensurate one , which has a three - sublattice ordering in the form of planar star - of - davids .
this lowest temperature commensurate cdw is predicted to be a metal in density functional theory calculations , but is seen to be an insulator in transport measurements ( and other probes ) .
it has long been believed that the insulator arises from a mott transition in the band at the fermi energy @xcite , but recent work has called this interpretation into question @xcite , as it is becoming clearer that the stacking of the orbital ordering along the c - axis also plays an important role , and could even be the origin of the insulating behavior .
the time - resolved arpes studies showed that the system has its gap collapse when it is driven and it also saw an interesting subgap resonance as the system relaxed back to equilibrium . an ultrafast core - level x - ray photoemission spectroscopy study measured the charge modulation order parameter of the material @xcite and found that it decreases , and then relaxes back , but never goes all the way to zero , even though the gap has collapsed .
more recently , it has been found that the system can relax to a metastable metallic nonequilibrium state @xcite that appears to be similar in many respects to the nearly commensurate cdw but with some clear differences .
electron diffraction studies found yet another nonequilibrium metastable phase @xcite .
other materials have also been studied .
tbte@xmath1 is another interesting material which has been examined with time resolved arpes @xcite . here
one can watch the closing and re - opening of the gap directly in momentum space .
in addition to seeing this behavior , two new results have emerged first , the photoemission rings at a frequency given by the ordering phonon in the system and second , the spectral gap can be reduced , but not all the way to zero , even if there is a high fluence and there are significant subgap states which have closed the gap .
the persistence of this spectral gap feature is something that naturally emerges in nesting - based cdws as we describe below .
finally , tise@xmath0 has been thoroughly studied @xcite .
this material is believed to be an excitonic insulator , which should respond fast to a pump because the ordering is electronic in nature and does not require phonons .
indeed , the response time is seen to be quite fast in this system ( under 50 femtoseconds ) , and a recent theory has been developed to describe the experiments @xcite .
it is clear that there is much more work needed to fully understand the behavior of these materials .
many - body theory has also seen a significant development in recent years .
dynamical mean - field theory @xcite ( dmft ) has emerged as one of the most useful methods for describing electron correlations in three - d materials .
its extension to nonequilibrium @xcite , enables a wide range of different many - body problems to be solved .
the theory can work both in the normal state and in commensurate ordered phases , and recently , there has been a range of work on the nonequilbrum properties of cdws .
this started with the exact solution of a simplified bandstructure model , where a range of different phenomena were studied : ( i ) pump / probe photoemission spectroscopy @xcite ; ( ii ) high - harmonic generation of light @xcite ; ( iii ) response to a large dc current and bloch oscillations @xcite ; and ( iv ) how quantum systems are excited as functions of the pump amplitude and frequency @xcite .
since then , this approach has been expanded to examine the behavior of nesting - driven electronic cdws as described by the falicov - kimball model @xcite and also being solved exactly @xcite . here , the bloch oscillations in response to a large dc field , and the time - resolved photoemission spectroscopy were both studied .
the cdw phase of the falicov - kimball model is quite interesting , because it displays an additional tricritical quantum - critical point , which is unique in that the order parameter is not suppressed to zero at the critical point
instead , the system transforms from an insulator to a metal @xcite .
we describe this unique phenomena in detail here .
this review is not intended to be an exhaustive review of the subject , instead , its intent is to carefully derive the needed formalism to solve these problems , explain how one controls the numerics to implement the solutions , and then discusses the properties of those solutions .
our focus has been entirely on our work , which represents nearly all of the work on these systems within the nonequilibrium dmft approach .
finally , we discuss how one can relate the results of the theory to experiment , and determine the origin of some of the features seen in these experiments .
most cdw order is complex in real materials and does not simply follow the peierls paradigm of an ab ordered phase @xcite .
nevertheless , the simplest case of an ab ordered phase provides a rich playground to examine the generic properties of many cdw systems @xcite .
it also is the easiest problem to solve , and hence we focus all of our efforts in this work to understanding the properties of such an ordered system .
we envision the lattice @xmath2 having @xmath3 sites and being bipartite , which means it consists of two disjoint sublattices : the a sublattice and the b sublattice , which are each connected by the hopping matrix @xmath4 .
the hopping only connects between sites in different sublattices and for standard systems like a simple - cubic lattice , or a square lattice , the ordering wavevector is @xmath5 , in which @xmath6 is the lattice spacing ; we work in units where @xmath7 here .
this allows us to express the formalism in two representative ways in real space or in momentum space .
due to the ordering , the translational symmetry is reduced in real space , requiring a lattice with a basis , while in momentum space , we have a coupling between momentum @xmath8 and @xmath9 for all momenta in the reduced brillouin zone , which is half the size of the original bz .
the ordering wavevector satisfies @xmath10 where @xmath11 is the position vector of the @xmath12th lattice site .
note that the bandstructure for a periodic lattice , where @xmath4 depends only on @xmath13 , can be expressed as @xmath14 the hamiltonian will be written in terms of spinless fermionic creation @xmath15 and annihilation @xmath16 operators for conduction electrons at site @xmath12 on the lattice .
these operators satisfy the conventional anticommutation relations @xmath17 and can be transformed to momentum space via @xmath18 in the ordered phase , the reduced brillouin zone ( rbz ) is defined via @xmath19 when there is only nearest - neighbor hopping and @xmath20 is restricted to the rbz .
every momentum @xmath20 is then coupled to the momentum @xmath21 which lies in the original bz , but outside the rbz . in momentum space ,
the two fermionic operators are denoted @xmath22 and @xmath23 , whereas , for the sublattice basis , we restrict the momentum summations to the respective sublattices , as follows : @xmath24 where the @xmath25 is required so that these fermionic operators satisfy the conventional anticommutation relations . with this notation , we then find that @xmath26 as we work on the development of the many - body formalism for pump / probe experiments on cdw systems , we will find it convenient to use both of these different bases to represent the fermionic creation and annihilation operators . converting between
the two is simple , with the unitary transformation that relates them satisfying @xmath27 we will be working with an all electronic model of the cdw , which can be described by the falicov - kimball model ( or a specific , noninteracting limit of the model described below ) @xcite , which has conduction electrons , denoted with a @xmath28 and localized electrons , denoted by an @xmath29 , which mutually interact .
the hamiltonian in an external electric field is @xmath30 where @xmath31 is the localized electron number operator ( which can be treated as a classical ising - like variable ) , @xmath32 is the conduction electron chemical potential , @xmath33 is the @xmath29-electron site energy and @xmath34 is the on - site coulomb repulsion . in the limit @xmath35 , and in the case of half - filling , where the density of conduction electrons and the density of localized electrons is each 0.5 , we find that the equilibrium solution has @xmath36 for @xmath37 and @xmath38 for @xmath39 .
the simplest model for a cdw fixes @xmath40 at those exact values for all @xmath41 . for the simplified model
, the system always has cdw order , and it can be solved by diagonalizing a bandstructure on a lattice with a basis ( although the dynamics are still complex in nonequilibrium ) . while in the falicov - kimball model , the asymmetry between the two sublattices decreases as the temperature increases until we reach @xmath42 , where the density of both particles becomes uniformly distributed , on average , throughout the entire lattice . in eq .
( [ eq : ham ] ) , the hopping matrix is time - dependent to model an electric field via the peierls substitution @xcite @xmath43 with the spatially uniform , but time - dependent electric field given by the negative of the time derivative of the spatially uniform vector potential : @xmath44 . in eq .
( [ eq : hopping ] ) , the hopping matrix @xmath4 is a constant spatially periodic matrix [ which we take to be nonzero only between nearest neighbors , where it is equal to @xmath45 , with @xmath46 the spatial dimension of the system ; we will work in units with @xmath47 and take the @xmath48 limit ] .
in other words , we will be working on the infinite - dimensional hypercubic lattice .
we re - express the kinetic - energy operator in terms of the bandstructure in eq .
( [ eq : bandstructure ] ) and the conduction - electron operators in the two different representations ( when in the ordered phase ) : @xmath49{\aunderbrace[l1r]{c_{\mathbf{k}a}^\dagger~c_{\mathbf{k}b}^\dagger } } \left ( \begin{array}{c c } 0 & \epsilon(\mathbf{k}-\mathbf{a}(t))\\ \epsilon(\mathbf{k}-\mathbf{a}(t ) ) & 0 \end{array } \right ) \left ( \begin{array}{c } c_{\mathbf{k}a}^{\phantom\dagger}\\ c_{\mathbf{k}b}^{\phantom\dagger } \end{array } \right ) , \label{eq : ke_ab}\\ & = & \sum_{\mathbf{k}\in \textrm{\footnotesize rbz } } \aoverbrace[l1r]{\aunderbrace[l1r]{c_{\mathbf{k}1}^\dagger~c_{\mathbf{k}2}^\dagger } } \left ( \begin{array}{c c } \epsilon(\mathbf{k}-\mathbf{a}(t))&0\\ 0&-\epsilon(\mathbf{k}-\mathbf{a}(t ) ) \end{array } \right ) \left ( \begin{array}{c } c_{\mathbf{k}1}^{\phantom\dagger}\\ c_{\mathbf{k}2}^{\phantom\dagger } \end{array } \right ) .
\label{eq : ke_12}\end{aligned}\ ] ] here , the bandstructure shifted by the vector potential is directed along the diagonal of the hypercubic lattice and thereby satisfies @xmath50 with @xmath51 and the field ( and vector potential ) oriented along the diagonal so that @xmath52 .
one can think of the second bandstructure @xmath53 as the projection of the velocity onto the direction of the electric field .
note how the @xmath54 matrix that represents the kinetic energy is off - diagonal in the sublattice representation and diagonal in the original momentum representation ( @xmath20 and @xmath21 ) . to a maximum time @xmath55 and back , ending with a spur parallel to the negative imaginary axis of length @xmath56.,scaledwidth=48.0% ] the many - body problem
is solved with contour - ordered green s functions which are defined on the kadanoff - baym - keldysh contour depicted in fig .
[ fig : keldysh ] @xcite .
the contour graphically illustrates the time evolution of the operators in the heisenberg representation , which evolve from the intial time ( @xmath57 ) to time @xmath58 , then from time @xmath58 to @xmath59 and finally from time @xmath59 back to the initial time , followed by an evolution along a segment of length @xmath56 parallel to the negative imaginary axis . the momentum - dependent contour - ordered green s function ( for @xmath20 in the rbz )
is defined by @xmath60 with @xmath61 and @xmath62 being the 1 , 2 or a , b subscripts depending upon the representation and @xmath63 $ ] being the partition function . the symbol @xmath64 is the time - ordering operator , which orders times _ along the contour_. the fermionic creation and annihilation operators are in the heisenberg representation where @xmath65 and the evolution operator satisfies @xmath66 where the time ordering is with respect to ordinary time . substituting the evolution operators into the definition of the green s function then yields @xmath67 for @xmath58 ahead of @xmath59 on the contour ( @xmath68 ) ; we employed the identities @xmath69 and @xmath70 .
one can now directly see why the contour runs from the initial time to @xmath58 , back to the initial time , and then along the imaginary axis ( if we think of the thermal factor as an evolution along the imaginary - time axis ) .
there are two green s functions that we can extract from the contour - ordered green s functions the retarded green s function ( which holds information about the quantum states ) and the lesser green s function ( which tells us how those states are occupied ) .
they are defined via @xmath71 and @xmath72 the self - energy is defined via the equation of motion . to begin
, we must first determine the noninteracting green s function .
this is found by setting @xmath73 in the hamiltonian . because the subsequent hamiltonian commutes with itself at different times
, the green s function can be found simply by determining the equation of motion for the fermionic creation and annihilation operators .
this yields ( with the integrals between @xmath59 and @xmath58 _ on the contour _ ) @xmath74}\rme^{-i\int_{t'}^td\bar t [ \epsilon(\mathbf{k}-\mathbf{a}(t))-\mu ] } & 0\\ \fl 0 & { \scriptstyle [ f(-\epsilon(\mathbf{k})-\mu)-\theta_c(t , t')]}\rme^{-i\int_{t'}^td\bar t [ -\epsilon(\mathbf{k}-\mathbf{a}(t))-\mu ] } \end{array } \right ) , \end{aligned}\ ] ] since @xmath75 .
we introduced the fermi - dirac distribution @xmath76 $ ] and the contour unit step function @xmath77 which is equal to 1 if @xmath68 and 0 if @xmath78 .
this is in the @xmath79 momentum representation , where the kinetic energy is diagonal .
the self - energy is diagonal in the @xmath80 sublattice representation , given by diagonal elements @xmath81 and @xmath82 ; it has no momentum dependence because we are solving the problem in dynamical mean - field theory , which has a local self - energy . converting to the @xmath79 representation yields @xmath83 & \frac{1}{2}[\sigma^c_a(t , t')-\sigma^c_b(t , t')]\\ \frac{1}{2}[\sigma^c_a(t , t')-\sigma^c_b(t , t ' ) ] & \frac{1}{2}[\sigma^c_a(t , t')+\sigma^c_b(t , t ' ) ] \end{array } \right ) .\ ] ] dyson s equation then yields the interacting green s function @xmath84^{-1}(t , t ' ) , \label{eq : dyson}\ ] ] which is the @xmath85 matrix element of the inverse of the operator inside the square brackets .
the inverse is with respect to both the time indices and the @xmath54 structure imposed by the ordered phase .
note that the green s functions and the self - energy are continuous matrix operators .
hence , they can not be easily evaluated numerically .
to do so , requires one to discretize the contour and approximate the operators by finite matrices , which can then be inverted using standard linear algebra methods .
the discretization is then extrapolated to zero to produce the approximation to the continuous matrix operator .
details for how to do this can be found in ref .
@xcite .
we are often interested in local quantities , such as the local green s function , which is found by summing the momentum - dependent green s function over all momentum in the rbz . in many cases
, it is more convenient to perform the summation over the entire bz , since each term appears twice in the summation and one does not need to trace over the final @xmath54 structure .
the difference is whether one sums over @xmath86 terms that are larger than zero .
if restricting to the rbz , then we must weight the @xmath87 terms by 0.5 , otherwise , those boundary terms will be over counted .
of course , because the self - energy is independent of momentum , the sum over momentum can be replaced by a two - dimensional integral over the joint density of states for @xmath86 and @xmath53 . in the @xmath48 limit
, we find that the joint density of states is a double gaussian given by @xcite @xmath88 if restricting to the rbz , then one needs to reweight the joint density of states , and in some cases perform a trace over the final @xmath54 matrix .
but , by convention , we often weight the green s functions so that the many - body density of states on the a sublattice and the b sublattice each have spectral weight of 1 . in that case , one averages over both to get the _ average _ local density of states .
the simplified cdw case is determined more easily than the full interacting case .
the self - energy simplifies to @xmath89 and @xmath90 .
then , the evolution operator becomes block - diagonal for each momentum ( or , equivalently for each @xmath86 , @xmath53 pair ) , which is described by a simple @xmath54 landau - zener - like system .
the full evolution operator is found by using the trotter formula for a given discretization @xmath91 and can be analytically found to satisfy @xmath92
\mathbb{i}\\ -i\left [ \epsilon\left ( \mathbf{k}-\mathbf{a}\left ( t+\frac{\delta t}{2}\right ) \right ) \sigma_z+\frac{u}{2}\sigma_x\right ] \frac{\sin\left [ \delta t\sqrt{\epsilon^2\left ( \mathbf{k}-\mathbf{a}\left ( t+\frac{\delta t}{2}\right ) \right ) + \frac{u^2}{4}}\right ] } { \sqrt{\epsilon^2\left ( \mathbf{k}-\mathbf{a}\left ( t+\frac{\delta t}{2}\right ) \right ) + \frac{u^2}{4 } } } , \nonumber\end{aligned}\ ] ] where @xmath93 is the @xmath54 identity matrix , @xmath94 and @xmath95 are the corresponding pauli spin matrices , and we employ a midpoint integration rule for the evaluation of the hamiltonian in the trotter factor .
the full evolution operator for this @xmath54 block then becomes @xmath96 and then this is repeated for each momentum point .
it turns out that the green s function for the simplified cdw model is determined entirely in terms of this evolution operator .
namely , we find that the retarded green s function in the 1,2 momentum representation is determined by just evolution operators between times @xmath58 and @xmath59 ( because the quantum states just depend on the instantaneous value of the hamiltonian ) via @xmath97 if we sum over momentum and convert to the @xmath80 representation , we find that @xmath98 , \ ] ] with the @xmath99 sign for the @xmath100 sublattice and the @xmath101 sign for the @xmath102 sublattice ; these green s functions are normalized so that @xmath103 .
note that the sum over momentum is replaced by a double integral over the two band energies weighted by the joint density of states .
the lesser green s functions are more complicated , because they depend on all times , not just the times between @xmath58 and @xmath59 .
this is reasonable , because how the states are occupied depends on the history of how the occupancy has evolved over time .
the final expression depends on the initial occupancies of the electrons .
these are given by the following @xmath104 here , we use the following notation @xmath105 . the final result for the lesser green s function in the 1,2 representation is cumbersome and is given by @xmath106,\end{aligned}\ ] ] @xmath107,\end{aligned}\ ] ] @xmath108,\end{aligned}\ ] ] and @xmath109.\end{aligned}\ ] ] the shortened symbol @xmath110 is employed in these equations . while the transformation to the @xmath80 representation is straightforward , the resulting equations are so long , that we do not write them down here .
however , it is important to calculate the order parameter of the conduction electrons , for both the falicov - kimball model and the simplified model .
it is given by @xmath111}=-\frac{\sum_{\mathbf{k}\in { \rm rbz}}\left [ g^<_{12}(\mathbf{k},t , t)+g^<_{21}(\mathbf{k},t , t)\right ] } { 2\sum_{\mathbf{k}\in { \rm rbz}}\left [ g_{11}^<(\mathbf{k},t , t)+g_{22}^<(\mathbf{k},t , t)\right ] } , \ ] ] which is bounded between 0 and 0.5 in equilibrium ( but can become negative in nonequilibrium ) .
similarly , the order parameter of the localized electrons is @xmath112 which is fixed at 0.5 for the simplified model , and reaches 0.5 at @xmath113 for the falicov - kimball model .
it is always nonegative , because it is fixed at its equilibrium value , and hence also has no time dependence .
while we have provided a complete solution of the simplified model , we have not yet described how one solves the dmft for the cdw state in nonequilibrium .
it is solved via an iterative algorithm , but we must work with matrices that have time discretized on the contour .
the algorithm starts with a guess for the self - energies on the two sublattices ( in the @xmath80 representation , the self - energy is diagonal in the @xmath114 space ) .
the iterative approach is then as follows : ( 1 ) for the given self - energies , compute the local green s function @xmath115 by summing dyson s equation in eq .
( [ eq : dyson ] ) over all momenta ( practically speaking , we use an integration over the joint density of states to do this and we do so in the @xmath80 representation ) ; ( 2 ) extract the effective medium @xmath116 from the local dyson s equation ( which has an additional @xmath54 matrix structure in the @xmath80 representation ) : @xmath117 ( 3 ) construct the ( diagonal ) impurity green s function from the diagonal components of the effective medium via @xmath118(t , t'),\ ] ] with @xmath119 or @xmath102 and @xmath120 with @xmath121 or @xmath102 ; ( 4 ) extract the self - energy for the impurity by solving @xmath122 ( 5 ) and finally setting the new self - energy for the lattice to equal that of the impurity and then iterates steps ( 1)(4 ) until it is converged
. one of the important checks is the short time behavior of the green s function .
it turns out that by carefully examining the definition of the green s function , one can find the coefficients of the taylor series expansion in relative time .
in particular , for the retarded green s function , when @xmath123 ( or @xmath124 ) , we immediately know that @xmath125 , because the anticommutator of two fermionic operators is equal to one ; this holds both in momentum space and in real space .
higher derivatives can be evaluated by taking commutators with the hamiltonian . remarkably , the first few derivatives do not depend on the field , so they hold in equilibrium and in nonequilibrium .
these results are also called moment sum rules , because that is what they look like when one converts the relative time to a frequency via fourier transformation .
since the first few moment sum rules ( or equivalently relative time derivatives ) of the green s function can be found exactly , they become an important tool in testing the accuracy of calculations @xcite . for the simplified cdw model
, one can evaluate a taylor series expansion of the time evolution operator and immediately find that the expressions given above satisfy the appropriate sum rules @xcite .
for the falicov - kimball model , we calculate the moment sum rules by numerically evaluating the first few derivatives of the retarded green s functions .
they serve as important numerical consistency checks of the approach , and are critical to ensure accuracy of the final results .
the derivation of these sum rules is straightforward , but tedious .
while one can do this both for the momentum - dependent green s functions and the local green s functions , we report them only for the local green s functions here . the standard way to report them is in terms of the many - body density of states for each sublattice , which is defined to be @xmath126 , with @xmath127 . here , the average time is @xmath128 and the relative time is @xmath129 .
the moments then satisfy @xmath130 which become @xmath131 @xmath132 and @xmath133 the expectation value @xmath134 measures the average density of the heavy particles on each sublattice ; this expectation value does not change with time .
when calculating momentum - dependent quantities , like the angle - resolved pes , one must be careful to work with gauge - invariant quantities to ensure that the object being measured is a true observable . in this work , we will focus on the angle - summed pes , or total pes , which , being a local quantity , is manifestly gauge invariant .
hence , we do not discuss gauge - invariance issues further here .
in addition , we work with a constant matrix element approximation . for a single band model in the normal state , a constant matrix element simply factors out of the pes expressions .
but when the system has multiple bands , so that the green s function is represented by a matrix ( here a @xmath54 matrix structure for the @xmath100 , @xmath102 sublattices ) , then the matrix elements can not be constant in all different bases they are constant in one basis , then they are related by a unitary transformation in another basis . while it is tempting to ignore this fact , and approximate the pes signal by the trace of the matrix green s function ( since it is an invariant ) multiplied by a constant , this only holds in the basis where the green s function matrix is diagonal ; the photoemission spectra may not satisfy positivity in this case .
otherwise , the pes ( and especially the angle - resolved pes ) can involve more complex contributions .
here we take the simplifying assumption that the pes is given by a constant matrix element multiplied by the sum of the local diagonal contributions in the @xmath100 , @xmath102 representation . the averaging over the two sublattices
is required because the probe pulse will uniformly irradiate both sublattices .
the result for the photoemission from each sublattice is then a two - time , probe - pulse - envelope - weighted , fourier transform , given by @xcite @xmath135 the symbol @xmath136 is the probe pulse envelope function , which we take to be a gaussian centered about the time @xmath137 : @xmath138 with @xmath139 the effective width of the probe pulse ( broader pulses mean more energy resolution , less time resolution and _ vice versa _ ) . for completeness , we discuss two other observables .
one is the current , which determines the time rate of change of the energy via @xmath140 and the other is the filling within each of the bands ( essentially the filling within bands with energy larger than zero or smaller than zero ) . for the current
, one finds @xmath141\left [ g^<_{\mathbf{k}11}(t , t)+g^<_{\mathbf{k}22}(t , t)\right ] , \ ] ] where @xmath142 is the particle band velocity and the green s function is in the 1,2 basis .
the result for the filling into the different bands has only been derived for the simplified model .
we refer to ref .
@xcite for those complete formulas . ,
red ) and a low amplitude pulse ( @xmath143 , green ) .
note how the sum rule is accurate within a few percent for the large amplitude case , but is quite poor for the smaller amplitude .
[ figure reprinted from @xcite , with permission],scaledwidth=60.0% ] , red ) and a low amplitude pulse ( @xmath143 , green ) .
note how the sum rule is again accurate within a few percent for the large amplitude case , but is quite poor for the smaller amplitude .
[ figure reprinted from @xcite , with permission],scaledwidth=60.0% ] we end this section with a discussion of how to perform the numerical calculations . in all cases , our goal is to determine the contour - ordered green s function or the retarded and lesser green s functions ( which can be extracted from the contour - ordered one ) .
the electric field is chosen to satisfy @xmath144 where @xmath145 is the magnitude of the field at time @xmath146 .
for the simplified model we have already shown that the green s function is directly found from the evolution operator , which decouples for each momentum .
furthermore , the retarded green s function is determined solely by the relative time , so it typically does not require much computation to evaluate it .
the lesser green s function knows about the previous history of the system , so it requires longer runs in time to determine it , since we must start from a time in the distant past before the field is turned on .
we have chosen , in this work , to evaluate the evolution operator via the trotter formula .
the only subtlety is how small of a time step do we take for each of the trotter factors .
this is then adjusted to ensure that the results have converged ( best to extrapolate to @xmath147 and use sum rules to verify the convergence )
. one of the benefits of this approach is that we maintain unitarity explicitly for the evolution operator because each trotter factor is determined analytically , and is manifestly unitary .
an alternative way to solve this problem is to employ a conventional differential equation solver .
the advantage of the differential equation approach is that they can be made adaptive to help ensure appropriate accuracy , but they often suffer from loss of unitarity for long runs over large time intervals , and hence are often less reliable than the trotter - based methods for these problems .
we employed the trotter approach for all results shown here .
sublattice for ( a ) equilibrium and ( b ) nonequilibrium at an average time well before the pump pulse ( @xmath148 ) .
this is for the simplified model of the cdw with @xmath149 , but also holds for the falicov - kimball model at @xmath113 .
( c ) because the retarded green s function has a long tail in the time domain , the density of states for a large range of average times is affected by the pump pulse .
inset , one can see the long tail of the green s function which decays like @xmath150 for the equilibrium green s function , perhaps somewhat faster for the nonequilibrium green s function .
[ figure reprinted from @xcite , with permission],scaledwidth=90.0% ] the falicov - kimball model calculations are more demanding , because they require the full nonequilibrium dmft algorithm .
we discretize the system , usually with @xmath151 , 0.05 , and 0.033 and then quadratically extrapolate to zero @xmath91 .
as a check on the accuracy , we compute the zeroth and the first moment sum rules ( the second moment accuracy is poorer during the initial part of the pulse ) .
empirically , we find that the equations converge more accurately when the amplitude of the pump pulse is large .
this is illustrated in figs .
[ fig : moments1 ] and [ fig : moments2 ] , where we plot the extrapolated moment sum rules for @xmath143 and @xmath152 , and find that the large - amplitude case has acceptable errors , while the other has too large errors to be useful .
it is surprising that this result is most accurate for the nonequilibrium regions ; when the system is in equilibrium ( left region of the figures ) , the errors are much larger .
for this reason , we work exclusively with @xmath152 in this work for the falicov - kimball model results .
we begin our discussion of the behavior of these systems by focusing on the density of states which will allow us to immediately discuss the phase diagram an quantum critical behavior . in both models
, one can show that the @xmath113 dos diverges as the inverse square root of the frequency at the band edges , which form the spectral band gap for the cdw .
this is illustrated in fig .
[ fig : cdw_dos ] ( a ) for the @xmath100 sublattice with the divergence at the upper band edge .
when we fourier transform this to relative time , the fourier transform has a long tail which decays like @xmath150 . for an accuracy of 0.1% in the dos
, one needs to run the relative time out to @xmath153 or more due to the long decay .
this extreme nonlocality in time can cause misconceptions when one is working with the nonequilibrium system . in particular ,
if we fix the average time and fourier transform with respect to the relative time , then once the relative time is large enough , one of the two times in the @xmath58 and @xmath59 basis will be earlier than the time when the field was applied and one will be later . hence even for large negative average times , the dos at that average time will be affected by the presence of the field , as depicted in fig .
[ fig : cdw_dos ] ( b ) and ( c ) .
when fourier transformed to frequency the dos has significant oscillations that occur due to the slope discontinuity when the field is turned on ( occurring near relative time of @xmath154 for this case ) .
this analysis can only be done for the simplified model , because the falicov - kimball model can not be calculated out to long enough times to see this behavior .
this behavior is generic , however , for the transient dos when the green s functions in real time have long tails in equilibrium ( in most cases these singularities , or sharp peaks disappear or are broadened in nonequilibrium , so the steady state dos with a field present often have shorter tails in time , but the long - tails return for the quasiequilibrium states after the pump is turned on for a pump / probe experiment ) . , which corresponds to a strongly correlated metal .
the curves correspond to different temperatures .
note how the singularity disappears at finite temperature and how the subgap states evolve .
inset is the order parameter for the corresponding dos , as indicated by the color .
[ figure reprinted from @xcite , with permission],scaledwidth=75.0% ] , which corresponds to the quantum critical cdw .
the curves correspond to different temperatures .
note how the singularity disappears at finite temperature and how the subgap states evolve .
inset is the order parameter for the corresponding dos , as indicated by the color .
[ figure reprinted from @xcite , with permission],scaledwidth=75.0% ] , which corresponds to the critical point for the mott insulator .
the curves correspond to different temperatures .
note how the singularity disappears at finite temperature and how the subgap states evolve .
inset is the order parameter for the corresponding dos , as indicated by the color .
[ figure reprinted from @xcite , with permission],scaledwidth=75.0% ] we next focus on the behavior of the falicov - kimball model in equilibrium at finite - temperature , because the electron correlations bring on a distinctive behavior which is quite different from the standard bardeen - cooper - schrieffer ( bcs ) paradigm @xcite .
in particular , if we define the spectral gap for the cdw to be the distance between the maxima that are the remnants of the divergence at @xmath155 in the @xmath113 dos , then we see that for most cases we will consider , the spectral gap remains fixed at @xmath34 all the way up until @xmath156 ( or quite close to @xmath42 ) .
this differs completely from the bcs paradigm , where the spectral gap is tied directly to the order parameter ( in that case the superconducting gap function ) , and the spectral gap shrinks as the order parameter shrinks until @xmath42 is reached where the spectral gap vanishes . here
, the phenomena is quite different .
we instead have two subgap minibands that form once @xmath157 , and they grow both in weight and in bandwidth until they close the subgap and then fill in the dos from below as the spectral gap features diminish from above and both meet at @xmath156 to produce the normal state dos . figures [ fig : cdw_dos_fk_metal ] , [ fig : cdw_dos_fk_crit ] , and [ fig : cdw_dos_fk_mott ] plot the dos in equilibrium for different temperatures for the metal , the quantum critical cdw and the critical mott insulator phases .
the different colors correspond to the different temperatures , with the dashed black line being the normal state .
inset are the order parameters for the light ( right ) and heavy ( left ) electrons .
note how the heavy electrons always have an order parameter that goes to 0.5 as @xmath35 , but the light electrons order parameter is always less than that , although it increases as @xmath34 increases .
but the behavior is even more complex than this .
the minibands initially start near the upper and lower band edges , but as @xmath34 increases , they migrate toward the center of the bandgap and they meet when @xmath158 .
this is the underlying quantum critical point for the cdw , because at this point , the system has a transition from a metal to an insulator at @xmath35 .
the metallic phase opens as a fan for higher @xmath41 as the dos becomes nonzero at the chemical potential and traces out the novel metallic cdw phase within the phase diagram .
this phase is most stable for temperatures below but near @xmath42 .
as @xmath34 is increased further ( up to @xmath159 ) we then have the mott transition . in this case , the minibands form and grow in weight and broaden as @xmath41 increases , but they never broaden to completely fill the gap .
instead the band edges stop at the mott insulator band edges , so the subgap region inside the mott gap never fills with any states .
once one is above @xmath42 , the dos becomes temperature independent , and either has nonzero weight at the chemical potential for the metal or has no dos at the chemical potential for the mott insulator .
, characterized by the opening of a gap in the single - particle density of states , which is the same on each sublattice . in the cdw ordered phase , where the densities of the particles are different on the two sublattices , there are three phases : ( 1 ) a weakly correlated cdw phase , which is continuously connected to @xmath73 ; ( 2 ) a strongly correlated cdw metal , which is present only at nonzero temperature , and emerges from the quantum critical point of the model at @xmath160 ; and ( 3 ) a strongly correlated cdw insulator , which is continuously connected to the mott insulator within the cdw phase .
the quantum critical point is nonstandard , because the order parameter for the cdw order varies continuously through the transition.,scaledwidth=65.0% ] the phase diagram showing all of these phases and their region of stability is plotted in fig .
[ fig : cdw_phase ] for the spinless falicov - kimball model on a hypercubic lattice .
the five different phases are indicated by the different colors and the boundary lines .
note how the quantum critical cdw region is extremely narrow for temperatures below 0.02 @xmath161 .
it becomes quite difficult to determine the precise phase boundary in this region because it is hard to calculate the dos accurately there .
the phase boundaries are more rigorously determined and the existence of the quantum critical point is explicitly proved on an infinite - coordination bethe lattice @xcite ( as opposed to the hypercubic lattice results shown here ) .
nevertheless , we work with the hypercubic lattice here , where the phase boundary is more challenging to explicitly determine .
note that this quantum critical point is rather unique .
an ordinary quantum critical point will occur at the terminus of a phase diagram where an order parameter is suppressed to zero . at that point
, there is a zero - temperature phase transition . in many quantum critical points ,
the original quantum critical region is hidden by a superconducting dome .
the case here is different for a number of reasons .
first , the spatial order parameter does not disappear at the quantum critical point , instead , it varies continuously at the critical point .
instead , the dos has a change of character as @xmath35 versus @xmath113 , the former being metallic and the latter being insulating .
nevertheless , there remains a strange metal fan above the quantum critical point , similar to more conventional quantum critical points . as a function of time for different pulse amplitudes ( zero is indicated by the magenta dashed line ) .
the purple dashed line is a zoom in of the black curve for long times .
the field is shown at the top in red .
[ adapted from @xcite , with permission],scaledwidth=55.0% ] now we turn to describing the conduction electron order parameter and how it varies with the pump pulse .
as electrons are driven by the field , they will flow through the material moving from the @xmath100 to the @xmath102 sublattice and being excited up to the higher energy band ( or de - excited to the lower energy band ) . as a result
, we anticipate that the conduction - electron order parameter will transiently change as a function of time .
this is indicated for the simplified cdw model in fig .
[ fig : cdw_non_order ] .
the cases with a pulse amplitude of 0.75 and 1 both oscillate and are pushed downwards , but settle into values that are positive for the long - time limit , while , when the amplitude is increased to 1.25 and 1.5 , the order parameter actually changes sign ! for larger amplitudes
we anticipate that the order parameter will continue to oscillate and the sign of the long - time limit may be difficult to determine without running through the whole calculation . in fig .
[ fig : cdw_fk_order ] , we show a similar figure , but this time for a large amplitude pulse ( @xmath152 ) in the falicov - kimball model with different initial temperatures .
the simplified cdw model will be closest to the low - temperature results .
initially , the system starts in equilibrium , and so the order parameter should be a constant , but the numerical results show some small dependence on time in this region ( on the order of a few percent ) which is an indication of the accuracy of the data after the extrapolations have been done .
inset , we show the total electron filling , which should be precisely 1 throughout the simulation .
its variation is another indication of the accuracy of these calculations , which are pushing the state - of - the - art to its limits .
the metallic case is in panel ( a ) and it shows rather expected behavior .
it starts off fairly flat , is reduced as the field is turned on , with large oscillations , and then settles into a rather flat result in the long - time limit , which is reduced from the original equilibrium value , but remains positive .
panel ( b ) of the quantum critical cdw is much more interesting .
it shows a nearly symmetric curve , where the order parameter is reduced , but in the end , the final value is quite similar to the initial equilibrium value .
we will see that this occurs because it is difficult to excite the quantum critical cdw .
finally , panel ( c ) shows the critical mott insulator where in all cases , regardless of the starting point , the order parameter is being suppressed almost to zero .
it is surprising that this suppression is more prominent for the mott insulator than it is for the metal , and there is no simple explanation for why this would be so . and different temperatures .
the convolution washes out a number of the features in the density of states , which will continue as the system is pumped .
panel ( e ) shows the same for the simplified model with @xmath162 and @xmath163 .
[ panels ( a - d ) reprinted from @xcite , with permission],title="fig:",scaledwidth=45.0% ] and different temperatures .
the convolution washes out a number of the features in the density of states , which will continue as the system is pumped .
panel ( e ) shows the same for the simplified model with @xmath162 and @xmath163 .
[ panels ( a - d ) reprinted from @xcite , with permission],title="fig:",scaledwidth=45.0% ] the formula for the time - resolved photoemission spectroscopy in eq .
( [ eq : pes ] ) involves a double - time fourier transform weighted by the probe envelope given in eq .
( [ eq : probe ] ) .
the form of the photoemission integral is that of a convolution .
when the probe width @xmath139 is narrow , one has high time resolution but poor spectral resolution and _ vice versa _ when the probe width is broad . to understand the effect this has on the features of the density of states , we show the convolution of the probe function used for the quantum critical cdw state of the falicov - kimball model , where @xmath164 with the local dos for different temperatures in fig . [ fig : dos_conv ] ( a d ) .
one immediately sees that the gap region is smoothed out , and the details of the subgap states are lost .
furthermore , as one approaches the normal state , the convolution becomes quite close to the original signal because it does not have any sharp features .
when the interaction is decreased to @xmath165 and the probe width is increased to @xmath163 , as shown for the @xmath113 data on the right , we see that the smoothing out of the features is much less and the gap is nearly fully formed .
the divergent peaks however remain smoothed out , because they require long tails in time before they are fully developed .
it is important to keep in mind the smoothed out features due to the convolution when we look at the time - resolved pes next , because those data will also have the smoothed out behavior . with @xmath166 and averaged over the a and b sublattices , for the simplified cdw model .
the electric field profile is shown above the plot .
[ reprinted from @xcite , with permission],scaledwidth=49.0% ] the results for the transient time - resolved pes for the simplified model with @xmath167 is shown in fig .
[ fig : pes_non ] .
the false - color plot has a logarithmic scale for the pes to emphasize the data at small values .
one can immediately see that the gap closes ( light blue region near @xmath146 ) when the field is present and then re - opens .
a substantial number of electrons are excited by the pulse , and there is a small band narrowing , but the spectral gap remains at about the same value for all times ( perhaps it shrinks by a percent or two ) . since the noninteracting system requires a field to be present for both excitation and de - excitation , the number of electrons excited into the upper band does not change once the pump pulse is over .
we will use the same false - color scale for all of the pes shown in this work . with @xmath168 , @xmath165 and averaged over the a and b sublattices , plotted in false color .
this system is a strongly correlated metal .
the electric field is shown above the plot .
the different panels are for the different temperatures in fig .
[ fig : cdw_dos_fk_metal ] with ( a ) the lowest and ( d ) the highest temperature .
( right ) waterfall image of vertical cuts through the tr - pes data plotted for different delay times and offset for clarity .
inset is the total integrated spectral weight .
only the data that conserves the spectral weight is shown .
this data corresponds to the case of panel ( a ) to the left .
[ reprinted and adapted from @xcite , with permission],title="fig:",scaledwidth=49.0% ] with @xmath168 , @xmath165 and averaged over the a and b sublattices , plotted in false color .
this system is a strongly correlated metal .
the electric field is shown above the plot .
the different panels are for the different temperatures in fig .
[ fig : cdw_dos_fk_metal ] with ( a ) the lowest and ( d ) the highest temperature .
( right ) waterfall image of vertical cuts through the tr - pes data plotted for different delay times and offset for clarity .
inset is the total integrated spectral weight .
only the data that conserves the spectral weight is shown .
this data corresponds to the case of panel ( a ) to the left .
[ reprinted and adapted from @xcite , with permission],title="fig:",scaledwidth=49.0% ] we now move on to the falicov - kimball model , where we are required to run the calculations with a much larger amplitude of the field , otherwise the extrapolated results have not yet converged .
the data given here all correspond to @xmath152 . in fig .
[ fig : pes_fk_metal ] , we show a series of false color images on the left and a waterfall plot for the lowest temperature on the right .
there are a few features to emphasize here which are different from what we saw in the simplified cdw case .
some of these may be arising from the fact that the field amplitude is so large now .
first , we see a significant narrowing of the band when the field is on ( notice the large blue region near @xmath146 ) .
because the total spectral weight is conserved , this narrowing comes with a sharpening of the peaks in the pes as well , which is a bit harder to see in the false color image , but may be clearer in the waterfall .
one can also clearly see that the spectral gap is shrinking , as the lower gap edge is being pushed toward 0 in the waterfall .
we inset the total spectral weight as a function of the probe delay , to show that the data we use has conserved spectral weight , while for too long or too short delays , our data becomes poor , and spectral weight is lost .
we do not show any data for the red points here .
the severe band narrowing is quite surprising , as it is coming from a dressing of the electronic states by the pump pulse .
the surprise is that the bandwidth is narrowed almost by a factor of two , which is a significant effect . with @xmath168 , @xmath160 and averaged over the a and b sublattices , plotted in false color .
this system is a strongly correlated metal .
the electric field is shown above the plot .
the different panels are for the different temperatures in fig .
[ fig : cdw_dos_fk_crit ] with ( a ) the lowest and ( d ) the highest temperature .
( right ) waterfall image of vertical cuts through the tr - pes data plotted for different delay times and offset for clarity .
inset is the total integrated spectral weight .
only the data that conserves the spectral weight is shown .
this data corresponds to the case of panel ( a ) to the left .
[ reprinted and adapted from @xcite , with permission],title="fig:",scaledwidth=49.0% ] with @xmath168 , @xmath160 and averaged over the a and b sublattices , plotted in false color .
this system is a strongly correlated metal .
the electric field is shown above the plot .
the different panels are for the different temperatures in fig .
[ fig : cdw_dos_fk_crit ] with ( a ) the lowest and ( d ) the highest temperature .
( right ) waterfall image of vertical cuts through the tr - pes data plotted for different delay times and offset for clarity .
inset is the total integrated spectral weight .
only the data that conserves the spectral weight is shown .
this data corresponds to the case of panel ( a ) to the left .
[ reprinted and adapted from @xcite , with permission],title="fig:",scaledwidth=49.0% ] as we move to fig .
[ fig : pes_fk_crit ] , we find even more surprising results . here , it is virtually impossible to excite the electrons at all into the upper band . at the lowest temperature ,
the number excited is almost equal to the number de - excited .
we continue to have the same band narrowing and peak sharpening due to field dressing of the states and the reduction of the spectral band gap for the cdw as before .
but now , the real hallmark is that the number of electrons excited to the upper band is small after the pump pulse is completed .
this compares reasonably to what we saw earlier for the order parameter , where we saw that it did not change significantly for the quantum critical cdw .
perhaps this behavior is tied to the existence of the conducting channel at the chemical potential in equilibrium ; if that conducting channel is efficient , it can lead to a channel for de - excitation which competes with the excitation and leads to a small net excitation . with @xmath168 , @xmath169 and averaged over the a and b sublattices , plotted in false color .
this system is a strongly correlated metal .
the electric field is shown above the plot .
the different panels are for the different temperatures in fig .
[ fig : cdw_dos_fk_mott ] with ( a ) the lowest and ( d ) the highest temperature .
( right ) waterfall image of vertical cuts through the tr - pes data plotted for different delay times and offset for clarity .
inset is the total integrated spectral weight .
only the data that conserves the spectral weight is shown .
this data corresponds to the case of panel ( a ) to the left .
[ reprinted and adapted from @xcite , with permission],title="fig:",scaledwidth=49.0% ] with @xmath168 , @xmath169 and averaged over the a and b sublattices , plotted in false color .
this system is a strongly correlated metal .
the electric field is shown above the plot .
the different panels are for the different temperatures in fig .
[ fig : cdw_dos_fk_mott ] with ( a ) the lowest and ( d ) the highest temperature .
( right ) waterfall image of vertical cuts through the tr - pes data plotted for different delay times and offset for clarity .
inset is the total integrated spectral weight .
only the data that conserves the spectral weight is shown .
this data corresponds to the case of panel ( a ) to the left . [
reprinted and adapted from @xcite , with permission],title="fig:",scaledwidth=49.0% ] in fig .
[ fig : pes_fk_mott ] , we show the same results for the critical mott insulator . here
, we see almost complete excitation in the system the weight in the upper and the lower bands appears to be nearly equal .
we continue to see the band narrowing and the spectral gap narrowing , which we saw before , but now , this system has quite strong excitation .
this result is surprising , because it excites more than the metal does .
perhaps , in this case , the de - excitation pathway suffers from some bottlenecks which make it less efficient . in any case
, the other curious feature is that the gap closure due to subgap states is much more modest here , and in fact , it closes with the initial and final fields , but reopens when the field amplitude is the largest ! a truly surprising result . in experiment ,
the most important features seen are the closing of a gap by filling in of subgap states , its reforming well after the pump , along with an oscillation of the pes that modulates at the same frequency as the phonon that is responsible for the cdw order . in addition , for the tbte@xmath1 experiment , as the fluence is turned up , the system has the spectral gap get reduced , but it never goes all the way to zero .
some of these features are captured in these simplified models of a cdw while others are not .
what these all electronic models see is a gap collapse by filling of subgap states , and a partial reduction of the spectral gap , but not all the way to zero .
they do not see the collapse of the gap for long times after the pump , nor do they see the long - time oscillations , both which are likely associated with the ordering phonon ( which is not part of this model ) .
the model calculations also show a decoupling of what is happening with the cdw order parameter , as measured by the modulation of the charge , and the spectral gaps , as measured in a pes experiment .
the model calculations also see some new phenomena which is yet to be seen in experiment .
this includes the preponderence of de - excitation and the difficulty with exciting a quantum critical cdw .
it also includes the large band narrowing and peak sharpening due to field dressing when the pump field is on .
what is next ? from the theory standpoint , the next step is to include the coupling to the phonons directly .
this should reproduce a number of the additional features seen in experiment , and will hopefully allow for a more complete solution of the problem . from the experimental standpoint , we hope that experiments might be done on materials that have there cdw transition driven by nesting , where these additional subgap features and quantum critical behavior may be present .
those systems could prove to be an interesting playground for novel physics . in this review , we covered a wide range of the theory needed to describe strongly correlated cdw materials in nonequilibrium .
we developed the theory both for the simplest model of a cdw , which is essentially a bandstructure with a basis , to describing the exact solution of an all electronic cdw described by the falicov - kimball model .
the latter having a unique quantum critical point within the cdw phase .
we used these formalisms to solve these problems for a wide array of different parameters .
technical reasons restricted the falicov - kimball model calculations to large amplitude fields only .
this work represents starting point for the theory of nonequilibrium pump / probe experiments in strongly correlated materials .
much more work needs to be done in the future to treat different kinds of order , the competition between different ordered phases , and different types of electron correlations .
in addition , extensions that include the possibility of incorporating more real materials properties , along the lines of density functional theory plus dynamical mean - field theory , but now in nonequilibrium , will also be important .
we hope to see and to participate in these developments in the coming years .
this work was supported by the department of energy , office of basic energy sciences , division of materials sciences and engineering under contract nos .
de - ac02 - 76sf00515 ( stanford / simes ) and de - fg02 - 08er46542 ( georgetown ) .
computational resources were provided by the national energy research scientific computing center supported by the department of energy , office of science , under contract no .
de - ac02 - 05ch11231 .
was also supported by the mcdevitt bequest at georgetown .
99 perfetti l , loukakos p a , lisowski m , bovensiepen u , berger h , biermann s , cornaglia p s , georges a and wolf m 2006 _ phys .
lett . _ * 97 * 067402 perfetti l , loukakos p a , lisowski m , bovensiepen u , wolf m , berger h , biermann s and georges a 2008 _ new j. phys . _ * 10 * 053019 hellmann s , beye m , sohrt c , rohwer t , sorgenfrei f , redlin h , kallne m , marczynski - bhlow m , hennies f , bauer m , fhlisch a , kipp l , wurth w and rossnagel k 2010 _ phys . rev
_ * 105 * 187401 stojchevska l , vaskivskyi i , mertelj t , kusar p , svetin d , brazovskii s and mihailovic d 2014 _ science _ * 344 * 177 vaskivskyi i , gospodaric j , brazovskii s , svetin d , sutar p , goreshnik e , mihailovic i a , mertelj t and mihailovic d 2015 _ sci . adv . _ * 1 * e1500168 han t - t , zhou f , malliakas c d , duxbury p m , mahanti s d , kanatzidis m g and ruan c y 2015 _ sci . adv . _ * 1 * e1400173 schmitt f , kirchmann p s , bovensiepen u , moore r g , rettig l , krenz m , chu j - h , ru n , perfetti l , lu d h , wolf m , fisher i r and shen z - x 2008 _ science _ * 321 * 1649 schmitt f , kirchmann p s , bovensiepen u , moore r g , j - h , lu d h , rettig l , wolf m , fisher i r and shen z - x 2011 _ new j. phys . _ * 13 * 063022 rohwer t , hellmann s , wiesenmayer m , sohrt c , stange a , slomski b , carr a , liu y , avila l m , kallne m , mathias s , kipp l , rossnagel k and bauer m 2011 _ nature ( london ) _ * 471 * 490 hellmann s , rohwer t , kallne m , hanff k , sohrt c , stange a , carr a , murnane m m , kapteyn h c , kipp l , bauer m and rossnagel k 2012 _ nature commun . _ * 3 * 1069 fazekas p and tosatti e 1979 _ philos . mag .
_ part b * 39 * 229 ; 1980 _ physica b + c _ * 99 * 183 darancet p , millis a j and marianetti c a 2014 _ phys
_ b * 90 * 045134 ritschel t , trinckauf j , koepernik k , bchner b , von zimmermann m , bergerh , joe y i , abbamonte p and geck j 2015 _ nat . phys .
_ * 11 * 328 gole d , werner p and eckstein m 2016 arxiv:1604.03784 metzner w and vollhardt d 1989 _ phys .
lett . _ * 62 * 324 brandt u and mielsch c 1989 _ z. phys .
b condens .
_ * 75 * 365 georges a and kotliar g 1992 _ phys .
_ b * 45 * 6479 freericks j k , turkowski v m and zlati v 2006 _ phys .
lett . _ * 97 * 266408 aoki h , tsuji n , eckstein m , kollar m , oka t and werner p 2014 _ rev
phys . _ * 86 * 779 shen w , ge y , liu a y , krishnamurthy h r , devereaux t p and freericks j k 2014 _ phys .
* 112 * 176404 shen w , kemper a f , devereaux t p and freericks j k 2014 _ phys .
_ b * 90 * 115113 shen w , devereaux t p and freericks j k 2014 _ phys .
235129 shen w , devereaux t p and freericks j k 2014 _ phys .
_ b * 90 * 195104 falicov l m and kimball j c 1969 _ phys .
* 997 matveev o p , shvaika a m , devereaux t p and freericks j k 2015 _ j. supercond . nov . magn . _ * 29 * , 581 matveev o p , shvaika a m , devereaux t p and freericks j k 2016 _ phys .
_ b * 93 * 045110 freericks j k , matveev o p , shvaika a m and devereaux t p 2016 _ proc .
spie 9835 ultrafast bandgap photonics _
98351f matveev o p , shvaika a m , devereaux t p and freericks j k 2016 arxiv : xxx hassan s r and krishnamurthy h r 2007 _ phys .
_ b * 76 * 205109 matveev o p , shvaika a m and freericks j k 2008 _ phys .
b * 77 * 035102 lemanski r and ziegler k 2014 _ phys .
_ b * 89 * 075104 johannes m d and mazin i
i 2008 _ phys .
_ b * 77 * 165135 peierls r e 1955 _ quantum theory of solids _
( oxford : clarendon press ) peierls r e 1933 _ z. phys . _ * 80 * 763 kadanoff l p and baym g 1926 _ quantum statistical mechanics _ ( new york : w a benjamin , inc . )
keldysh l v 1964 _ zh .
fiz . _ * 47 * 1515 [ 1965 _ sov .
. jetp _ * 20 * 1018 ] freericks j k 2008 _ phys .
_ b * 77 * 075109 turkowski v and freericks j k 2005 _ phys .
_ b * 71 * 085104 freericks j k and turkowski v 2009 _ phys .
_ b * 80 * 115119 ; 2010 _ phys .
_ b * 82 * 129902(e ) freericks j k , krishnamurthy h r and prushke t 2009 _ phys .
lett . _ * 102 * 136401 bardeen j , cooper l n and schrieffer j r 1957 _ phys . rev . _ * 108 * 1175 | in this review , we develop the formalism employed to describe charge - density - wave insulators in pump / probe experiments using ultra short driving pulses of light . the theory emphasizes exact results in the simplest model for a charge - density wave insulator ( given by a noninteracting systems with two bands and a gap ) and by employing nonequilibrium dynamical mean - field theory to solve the falicov - kimball model in its ordered phase .
we show both how to develop the formalism and how the solutions behave .
care is taken to describe the details behind these calculations and to show how to verify their accuracy via sum - rule constraints .
_ keywords _ : pump / probe experiment , charge - density - wave insulator , time - resolved photoemission spectroscopy , nonequilibrium dynamical mean - field theory |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.